Jump to content
retro_doog

New power supply project for TI-99/4a

Recommended Posts

3 hours ago, Kchula-Rrit said:

When I saw the part about "burning off 7V" I just had to comment that, after a year of service, one of the regulators on my console died of 'heat stroke' after giving out a piercing shriek.  An autopsy confirmed that one of the voltages had gone.  That panel next to the cartridge go hot enough to hurt my fingers.  No computer should die that way.

 

Anyway, I went down to the surplus store and found a +12,+/-5V brick to replace the TI one, and just wired it directly to the motherboard.  It's worked great ever since.  I'd like to find a +12V to a three-output converter so I can run mine off a car battery during power outages.

 

As far as the power switch goes, I got a four-pole toggle switch to switch the brick's output.  I'm not a fan of the current trend of having everything controlled by the computer, just too complicated and adds expense.  Also, when the computer goes "haywire" or locks up it's just so satisfying to throw a switch, rather than just holding a d*mn button for several seconds and hoping it's not too late.

 

Also, why don't computers have reset-buttons any more?  They think programs never lock up?

 

K-R.

 

The "burning off" is simply just the way linear voltage regulators work. Linear regulators are lower cost and take up much less board area than switching regulators, and create a much "cleaner" rail. However, creating 5V from 12V with a linear is a huge waste of power. In modern designs, when a noise free supply mandates use of a linear regulator for sensitive analog measurements or audio use, modern LDO or Low dropout voltage regulators are used. So, so to get from 12V to a clean 5V, you first have a switching regulator reduce the 12V to 6 or 7V then apply that to a low dropout linear regulator to get a clean 5V. Now only 1V or 2V worth of power is being wasted. Of course, the amount of power actually wasted depends on the current draw as well. It can be appropriate to create a 5V "trickle" rail from 12V with a linear regulator if it is only consuming <100mA to drive a realtime clock or system management unit(for things like soft power).

 

Modern computers and modern operating systems need more than just a hardware reset switch or you could create many ways to "brick" your system, or at least corrupt it to the point of needing to be re-formatted/installed. These old 8-bit systems had few power rails, less asynchronous clock domains, and no caching. When you wrote to the disk, the blocks were written when commanded and contrl returned to the program only after the write was complete for example. Modern hardware and operating systems use multiple levels of caching which is local higher speed storage that is only periodically read/written to/from the main source/destination. This applies to main memory in the case of the processor caches and to mass storage in the case of your file system. If you were to pull the plug(a hard reset is no different) anything in the caches would not get to their final destination and could easily leave things in a corrupt state. This is even more critical with SSDs as they constantly need to shuffle logical blocks around as part of their wear leveling and the also auto encrypt all data. Without proper power sequencing, it would be very easy to actually brick the SSD into not even being re-formatable at the OS level.

 

Ah, power sequencing! Our old machines only had a few power rails. Usually +5V for the digital logic, +12V generally for analog stuff like video/audio/disk drive motors or anything else that needed power, and in the older generation machines of which the TI sits on the border, a negative rail to bias old NMOS and early CMOS digital ICs. Not too long after the early 80's machines only needed one digital rail, the +5V. Then in order to facilitate faster clock speeds, lower power consumption, and to accommodate smaller transistor processes, all of which go hand in hand, the "digital rail" was reduced from +5V to +3.3V to +2.5V, +1.8V, +1.0V, and in todays fastest processors and memory, even sub 1V logic levels. A modern computer actually runs on a mix of ALL of the above voltages I listed at the same time! Moreover most ICs will have a lower "core voltage" so the internal logic can run faster and use less power and have a higher "I/O rail" to talk to other ICs on the MLB. All of these mixed voltage levels created the need for special power sequencing, where the rails are brought up in a proper timing relative to other rails. All of this plus other functions like thermal handling are done be another small processor(usually a microcontroller) generally referred to as the "system management unit" or SMU. It orchestrates the sleep/wake/hibernate, powerup/power down, and warm/cold reset as well as recovery from hard power losses(auto reboot for servers and the like). All this to try to ensure you can't/won't corrupt or brick or even destroy your logic IC's by typing the wrong thing or pushing the wrong button. There's just no way to accommodate a hard reset button on modern machines.

 

As far as programs locking up, modern operating systems are very adept at recovering from an individual program or process that goes awry. Only a hard crash of the OS itself would require a forces reset of the machine. I'm a Mac guy, so let me illustrate: On an Apple II, if a program crashed your whole computer froze up or if you were lucky dumped you into the "monitor" and you had to hard or soft reboot(power switch, ctrl-reset, or PR#6). Later on the early Macs, a program crash would give you the system bomb and a dialog box with a button to click to reset the machine. Here you can see the operating system geting smart enough to try and salvage what was in progress but still needing a warm reset directed by the user. You could still get hard crashes that required the reset button, but as the OS progressed through System 6, 7, and 8 hard crashes became less common. On to OS-X, and now a program can crash, but leave you more or less "cleanly" back at the finder. In fact, the other programs you had running didn't lose any data or crash themselves! You could still get a Kernel Panic and they were not uncommon in the early days/years of OS-X. Fast forward to today, I'm running MacOS 10.13 "High Sierra" and I can't recall having a kernel panic for MANY years. Probably not since before 10.6. Maybe in 10.4 through 10.8, I'd need to reboot the machine every week or two to give it it's "pep" back. For the last 5-6 years at least though, I can let my machine run for months or even over a year, only needing to reboot it for certain OS updates or again to give it it's pep back. If these machines had hard reset buttons to the processor, they would be in much worse shape. There are literally a thousand things going on in your modern computer besides the program you happen to be running and yanking those in flight would wreak havoc.

 

Sorry for the long post. I just thought I'd give a treatise on why things aren't as simple as they were back in the old console days...

Share this post


Link to post
Share on other sites
20 minutes ago, HOME AUTOMATION said:

Yikes! With only four 150uF "bulk" caps on that supply, I'm not sure that would be sufficient for the TI Console. Despite being rated for 160W(and most of that capacity is probably on the 3.3V rail which the TI can't even use), the power architecture of the ITX machines these are designed for is not the same as our TI consoles.

 

Anyway, This is the reason I'm creating the board described in this thread! To make a drop-in, form factor correct, non-hacky, power converter replacement that meets the specs for THIS machine and will allow for a range of simple DC voltage input adapters. 😇

 

Edited by retro_doog
Gramma'

Share this post


Link to post
Share on other sites
13 minutes ago, HOME AUTOMATION said:

TI transformer input power: 40W...

 

:ponder:

My TI wall adapter(not the floor one with the extra fuse) doesn't have an input power rating, but it has the output power as: 18.0VAC 22.0VA ; 7.5VAC 1.0VA.

 

A volt-amp is basically a Watt, so the output power is 23Watts. Oh wait. The input amperage is 0.3A so at 120V, yes that's 36W! Maybe the floor bricks use a bit more. That means that even the transformer is burning power. That makes sense, the transformer is a pair of coils - the old transformers always "Buzz" and have vibration, that's actually mechanical energy with converts to heat as well and is not delivered to the outputs. So, even more reason to design something that uses both a modern switching type AC/DC at the wall + modern high efficiency switching DC/DC regulators inside the console. Maybe I can make our TI-99's "Energy Star" rated? :P

 

Edited by retro_doog
Type-O
  • Like 1

Share this post


Link to post
Share on other sites
2 hours ago, HOME AUTOMATION said:

Yeah. The 12vdc issue is a little different.

 

I think the best thing... to protect the old parts on TI boards is... a resistive power supply!:)

I'm pretty sure I can find a way to make this supply run on 12VDC Input. Especially since the 12V rail usually has a higher tolerance than he 5V rail. I've seen +/-10% specified for some older computers 12V supplies, so we could be as low as 10.8V and still have no operating problems. With an LDO for the 12V rail at the typical low current draw I'm seeing, I could probably supply a well regulated 11.7V or so which should be more than fine. One should be able to power their TI-99 from a car battery with the board I'm designing.

 

Not sure what you mean by resistive power supply. Can you elaborate on what that is and how it would protect older components on the main board?

 

Share this post


Link to post
Share on other sites

Heat is one of the main enemy of our old machines. I would go with an external solution if I could. Something that can provide the required voltages in a clean filtered way and with enough current to power the 32k and TiPi from the expansion bus.

Share this post


Link to post
Share on other sites
27 minutes ago, Badaboom said:

Heat is one of the main enemy of our old machines. I would go with an external solution if I could. Something that can provide the required voltages in a clean filtered way and with enough current to power the 32k and TiPi from the expansion bus.

Heat won't be a problem with the board I'm designing. Modern switching supplies can be 90+% efficient. Any fully external solution that provides clean rails will be a brick. Any modern 32K expansion should draw less than 100mA and the TiPi is probably less than 500mA. I have two candidates for the 5V switching supply, one is rated for 2A and one for 3A, the console+parsec only draws 1A, so we'l have plenty of margin there...

Share this post


Link to post
Share on other sites
3 hours ago, retro_doog said:

Not sure what you mean by resistive power supply. Can you elaborate on what that is and how it would protect older components on the main board?


Well, of course all P/Ss have some resistance, causing voltage to drop somewhat, under loading.:twisted: Regulators counter this, by effectively lowering the resistance between their inputs and outputs. This supplies ...what the circuit demands ...and can supply more... perhaps at times, undesirably.:evil:

 

Discrete semiconductor components, designed to drive many more inputs(for example) or operate at higher speeds, than are employed in the /4A 's design,:thumbsup: are not best utilized in the long term, by being constantly provided with enough current to run at full specs.:thumbsdown:

 

When the P/S's resistance is higher|:)(generally speaking), it takes longer for an overload(burnout) to occur ...should something go wrong!:o

 

Under best conditions, the board's distribution, and power supply, will not supply enough sustained current to cause a part(s) to fail, even when shorted!:cool:

 

Determining the best resistance and voltage range allowances, can be accomplished through design considerations:ponder: or extensive aftermarket fault testing.:grin:

 

:)

Share this post


Link to post
Share on other sites
2 hours ago, HOME AUTOMATION said:


Well, of course all P/Ss have some resistance, causing voltage to drop somewhat, under loading.:twisted: Regulators counter this, by effectively lowering the resistance between their inputs and outputs. This supplies ...what the circuit demands ...and can supply more... perhaps at times, undesirably.:evil:

 

Discrete semiconductor components, designed to drive many more inputs(for example) or operate at higher speeds, than are employed in the /4A 's design,:thumbsup: are not best utilized in the long term, by being constantly provided with enough current to run at full specs.:thumbsdown:

 

When the P/S's resistance is higher|:)(generally speaking), it takes longer for an overload(burnout) to occur ...should something go wrong!:o

 

Under best conditions, the board's distribution, and power supply, will not supply enough sustained current to cause a part(s) to fail, even when shorted!:cool:

 

Determining the best resistance and voltage range allowances, can be accomplished through design considerations:ponder: or extensive aftermarket fault testing.:grin:

 

:)

Well, I had written a very long winded response, but it didn't stick. I guess that's the forum's way of telling me to be less verbose :P

 

I'll give a much shorter answer this time and invite any follow up questions:

 

• Despite the term power "supply", it is the load that determines how much power is "drawn". It's not possible for the supply to "force" too much power into the board. If the load circuit has no faults(shorted capacitor for instance), then it will draw only what it needs.

 

• There is no danger in interfacing modern silicon with stronger output drivers or faster(shorter) propagation delays/speeds from an electrical standpoint. The load receiving circuit will only draw what it "needs". Faster logic may cause improper timing operation if the original design is trying to "cleverly" exploit known propagation delays(ie chain of buffers for delay or feedback inverters for oscillators). That would be more likely to just cause improper operation than circuit damage, though. There can be exceptions of course(bus contention/overlap due to higher speeds).

 

• If the power supply voltage is drooping under load(like a resistive supply would) that actually CAN damage circuitry. There are some cases where the load requires fixed power where if the voltage drops, more current is drawn which can exceed resistor power dissipation, or Diode/Transistor maximum current. Also voltage drops in digital circuitry will decrease the signal to noise ratio and result in bit errors/flaky behavior. One of the reasons a line level brown out damages some electronic equipment is because as the voltage drops, current increases to draw constant power.

 

• The best way to prevent overloads is to put fuses or active current limiting circuits on the supply outputs! I plan to put Poly Thermal Fuses on my supply design. They "auto reset" when they cool down.

 

• The supply Im designing, while having significant overhead to accommodate some external devices will likely deliver less max current in a short circuit condition than the OEM supply, so it should be much more reliable and even safer for our aging consoles. I'll probably do some "destructive" testing on an OEM unit as part of my research...

 

Hope this helps explain some of how this stuff works. If anyone wants, I can give a primer on the differences between linear supply operation and switching supply operation 😎

  • Like 2

Share this post


Link to post
Share on other sites
13 minutes ago, retro_doog said:

There are some cases where the load requires fixed power where if the voltage drops, more current is drawn which can exceed resistor power dissipation, or Diode/Transistor maximum current.

Hmm, Is the regulator, the chicken or the egg?:twisted:

Even this scenario CAN be compartmentalized in situ. I believe.

 

    P.S.

Admittedly the type of P/S I suggest is somewhat unique, and might otherwise, fail to perform, in some situations. Particularly with additional loads!

 

    P.P.S. Your plan does sound ...well considered.:ponder:

Share this post


Link to post
Share on other sites
24 minutes ago, HOME AUTOMATION said:

Hmm, Is the regulator, the chicken or the egg?:twisted:

Even this scenario CAN be compartmentalized in situ. I believe.

 

    P.S.

Admittedly the type of P/S I suggest is somewhat unique, and might otherwise, fail to perform, in some situations. Particularly with additional loads!

 

    P.P.S. Your plan does sound ...well considered.:ponder:

I did look up resistive and capacitive "transformer-less" supplies and what I found were worse than either liner or switching supplies. They are horribly inefficient and suitable only for very small loads before significant voltage drops. I believe this explains a lot of very small but super cheap wall supplies I've come across...

 

But again, to the regulator the load is just some perceived impedance(the more dynamic term for resistance outside of static resistive only loads). Ohms law dictates that the supply cannot "push" extra power into the load outside of a critical component failure, so the some fault or failure would have to precede the supply delivering more power than normally needed, which could, of course create additional component failures in either the load circuit or the power supply.

 

Still, even in the short circuit case, no power supply can supply infinite current. Many supply designs will just "fall flat" and not necessarily create smoke and fire. ;)

 

Virtually all modern regulator ICs have built in fault protection usually in the form of thermal shut down as well as often having over-voltage and under-voltage lockouts. The ones i plan to use are probably only going to have thermal protection, but the PTC fuses should add some additional protection. In fact they are designed similar to your resistive supply desired behavior. When the current exceeds a threshold, they heat up which raises their normally near zero resistance, thereby limiting current flow. Except unlike the resistive supply, under the normal range of current draw they have no virtually resistance/voltage drop. If you have a continuous fault, you have other problems, of course, but both the load board and the power supply are protected from additional damage.

Share this post


Link to post
Share on other sites
10 hours ago, retro_doog said:

Heat won't be a problem with the board I'm designing. Modern switching supplies can be 90+% efficient. Any fully external solution that provides clean rails will be a brick. Any modern 32K expansion should draw less than 100mA and the TiPi is probably less than 500mA. I have two candidates for the 5V switching supply, one is rated for 2A and one for 3A, the console+parsec only draws 1A, so we'l have plenty of margin there...

Oh, I wasn't arguing against your project, I was only talking about what "I" would like to do.

 

 

Share this post


Link to post
Share on other sites
On 7/8/2020 at 3:59 AM, Badaboom said:

Oh, I wasn't arguing against your project, I was only talking about what "I" would like to do.

 

 

Oh sure, I wasn't reading your commentary in a negative way. I just believe that the solution with the broadest appeal(as broad as can be for these old machines) is a drop in solution which works with a large variety of external supplies. Some users, like console + cart only, will be able to get by with a very small, low power wall adapter, but the internal supplies wil be able to handle reasonably higher loads provided a larger wattage external adapter is paired with it. That's the beauty of switching power supplies, they are very adaptable as long as you don't exceed the maximum output power. Some will be quite a bit less efficient at very low power draw or will go into power saving modes that are noisier, but we don't have to worry about that as our machines don't have any dynamic power saving modes like modern machines, so these supplies will be running at very high efficiencies and will only generate a fraction of the heat that the original linear supplies do...

  • Like 1

Share this post


Link to post
Share on other sites

OK I've some some more research including looking at the scanned document with the TI Console Technical Requirements and here's what I found:

 

The +12V does not appear to leave the console - It goes to the TMS9900, The DRAM Chips, Some pullup network on the GROM Bus(looks like precharge bus parking for the normally open collector data bus) and a clock generator chip. So... It would appear that I really do not need to provide any additional power capacity beyond the 250mA I originally measured. Also, the tolerance appears to be +/-10%, so that rail could be as low as 10.8V and the components would operate normally. This means I could use a very Low drop out voltage regulator with a 12V input supply and as long as the LDO output is a tighter tolerance operation would be satisfactory. I still may also add an option for a low current boost regulator to see if really small power plug type wall adapters can work(at least on the main console). I may be able to use a voltage doubler or tripler at this low current draw followed by an LDO or just a filter. I'll play around with some circuits to see...

 

On the +/-5V rails at the side I/O connector. Those are documented as being meant ONLY for the speech synthesizer and spec 50mA each Max. The PEB clearly has it's own hefty internal supply and does not draw from these. Even the circuitry on the flex cable derive their power from the +8 that comes from the PEB. I'm guessing the many "Sidecar" expansion boxes offered had their own wall supplies or power cords as you can't run much off of 50mA, especially in the TTL and 5V CMOS days. So... My plan to roughly double the capacity of the low voltage rails from 1A to 2A on the +5 and from 100mA to 200mA on the -5 should be more than enough. In fact there may or may not be trace width constraints on the PCB traces that prevent drawing too much additional power anyway! There will be more than enough power to accommodate anything that has already been designed to draw power from the side port and maybe even allow for devices that previously required their own wall adapters to run from console power. I'd probably limit that to the <500mA that a R-Pi draws, though.

 

On top of that, I've decided for the sake of simplicity not to add any fancy "soft power" features or the like. Anything that requires hacking wires out of the box or onto the other circuits in the console is an anti-goal. 99.4a% of the people are going to be just fine flicking a mechanical switch like they always did in exchange for a perfectly fitting drop in board with zero soldering required. I will have some sort of LED color select-ability though. Perhaps just jumpers or a DIP switch to "mix" colors allowing for 7 colors? Something fun and unique so that from the outside you know you have a little something special under the hood. 😎

  • Like 2

Share this post


Link to post
Share on other sites
7 hours ago, retro_doog said:

OK I've some some more research including looking at the scanned document with the TI Console Technical Requirements and here's what I found:

 

The +12V does not appear to leave the console - It goes to the TMS9900, The DRAM Chips, Some pullup network on the GROM Bus(looks like precharge bus parking for the normally open collector data bus) and a clock generator chip. So... It would appear that I really do not need to provide any additional power capacity beyond the 250mA I originally measured. Also, the tolerance appears to be +/-10%, so that rail could be as low as 10.8V and the components would operate normally. This means I could use a very Low drop out voltage regulator with a 12V input supply and as long as the LDO output is a tighter tolerance operation would be satisfactory. I still may also add an option for a low current boost regulator to see if really small power plug type wall adapters can work(at least on the main console). I may be able to use a voltage doubler or tripler at this low current draw followed by an LDO or just a filter. I'll play around with some circuits to see...

 

On the +/-5V rails at the side I/O connector. Those are documented as being meant ONLY for the speech synthesizer and spec 50mA each Max. The PEB clearly has it's own hefty internal supply and does not draw from these. Even the circuitry on the flex cable derive their power from the +8 that comes from the PEB. I'm guessing the many "Sidecar" expansion boxes offered had their own wall supplies or power cords as you can't run much off of 50mA, especially in the TTL and 5V CMOS days. So... My plan to roughly double the capacity of the low voltage rails from 1A to 2A on the +5 and from 100mA to 200mA on the -5 should be more than enough. In fact there may or may not be trace width constraints on the PCB traces that prevent drawing too much additional power anyway! There will be more than enough power to accommodate anything that has already been designed to draw power from the side port and maybe even allow for devices that previously required their own wall adapters to run from console power. I'd probably limit that to the <500mA that a R-Pi draws, though.

 

On top of that, I've decided for the sake of simplicity not to add any fancy "soft power" features or the like. Anything that requires hacking wires out of the box or onto the other circuits in the console is an anti-goal. 99.4a% of the people are going to be just fine flicking a mechanical switch like they always did in exchange for a perfectly fitting drop in board with zero soldering required. I will have some sort of LED color select-ability though. Perhaps just jumpers or a DIP switch to "mix" colors allowing for 7 colors? Something fun and unique so that from the outside you know you have a little something special under the hood. 😎

Isn't +12 used by the tv modulator and thus present on one pin of the video out?

  • Like 1

Share this post


Link to post
Share on other sites

At the risk of suggesting feature creep, I'd like to see some header pins attached so a USB charging port could be installed in the TI, and current requirements compensated for.  An extra useful feature could help drive sales.

 

If we have to open up the TI anyway...

 

 

usb port.JPG

Share this post


Link to post
Share on other sites

IMHO, as an option, I'd like to just make a board for the Meanwell that lets you:

 

* Hook the female 5 pin DIN connector with a PCB that fits the PSU hole on the back of the console

* Make a board that fits on the inside of the console with a switch.  We would have to figure out a circuit that handles triple pole switch that fits into the switch in the console *or* a 5V triple pole relay that the switch controls.   That's all that would really be on this PCB.

 

This would be insanely cheap.  You're only paying for a BOM of:

* 5 pin female DIN

* PCB rectangle board for the DIN

* Four wires from female DIN PCB to switch PCB

* Switch PCB with either:

** Triple pole switch -or-

** Triple pole relay and single pole switch

* Four wires and connector to motherboard

 

And of course, the Meanwell PSU.  

 

I haven't specced out anything yet.  But the triple pole equipment is bound to be fun to find.

 

 

  • Like 2

Share this post


Link to post
Share on other sites
9 hours ago, Badaboom said:

Isn't +12 used by the tv modulator and thus present on one pin of the video out?

Indeed it appears it does! That was hiding near the middle of the schematic page where the scanner misses the binding edges of the pages in the middle... I don't own a modulator to test, but I suspect it does not draw a significant additional load, and the attach rate is going to be pretty low on these. I'm guessing the "purist" crowd that still attaches these via the RF modulator is going to prefer the original TI power brick as well...

Share this post


Link to post
Share on other sites
8 hours ago, Omega-TI said:

At the risk of suggesting feature creep, I'd like to see some header pins attached so a USB charging port could be installed in the TI, and current requirements compensated for.  An extra useful feature could help drive sales.

 

If we have to open up the TI anyway...

 

 

usb port.JPG

Doing my best Walter Skinner(imagine the words mumbled and slurred together): "for-the-purpose-of?..."

 

In all seriousness, I don't see this as being something that most would want. There's a sizable gap between, "while my machine is open anyway" and "I think I'll hack a hole in the side to mount a panel mounted USB power port...

 

Still, I'm sure I'll have test headers on the board for all rails, so if someone wants to add additional harni (Plural for harness? :P) they can knock themselves out. I'm still going to fix the max 5V at 2.0A, though. That's plenty for the console's current(ha! a pun!) needs plus 500mA for the side port and 500mA I guess for a USB port like you want above...

  • Thanks 1

Share this post


Link to post
Share on other sites
14 minutes ago, retro_doog said:

Doing my best Walter Skinner(imagine the words mumbled and slurred together): "for-the-purpose-of?..."

 

A USB port on the TI could be used for many things, just off the top of my head...

 

1) If it was UNSWITCHED as dedicated close by power tap for the Pi ZERO on the TIPI/32K so we would not have to shut down power all the time.

2) To power for our cellphones as we are using them as a WiFi hub for our TIPI's to call into the BBS, check the weather, or doing anything at MYTI99.COM.

3) IF UNSWITCHED, as a convenient place to charge our cell phones at home.

4) For a nightlight to aim at the keyboard for those late night sessions.

5) The next big thing.TM

Share this post


Link to post
Share on other sites
8 hours ago, acadiel said:

IMHO, as an option, I'd like to just make a board for the Meanwell that lets you:

 

* Hook the female 5 pin DIN connector with a PCB that fits the PSU hole on the back of the console

* Make a board that fits on the inside of the console with a switch.  We would have to figure out a circuit that handles triple pole switch that fits into the switch in the console *or* a 5V triple pole relay that the switch controls.   That's all that would really be on this PCB.

 

This would be insanely cheap.  You're only paying for a BOM of:

* 5 pin female DIN

* PCB rectangle board for the DIN

* Four wires from female DIN PCB to switch PCB

* Switch PCB with either:

** Triple pole switch -or-

** Triple pole relay and single pole switch

* Four wires and connector to motherboard

 

And of course, the Meanwell PSU.  

 

I haven't specced out anything yet.  But the triple pole equipment is bound to be fun to find.

 

 

Well, I have several personal goals that this method defies:

 

• A 5-pin DIN Male Plug Shell is slightly bigger than the rectangular opening in the back for power, so the receptacle would either need to "poke out" of the back of the unit, or you's have to do some filing to enlarge the opening. I say "Yuck!" to both :P

 

• The board for the switch or switch relay would need to be about as large as the original supply board due to the mounting screw spacing, and the fact the on Beige units, only two screws are affixed and two bosses hold the board somewhat stable in order to make sure the internal power switch does not become decoupled from the enclosure switch actuator. (Still need to look inside my Black and Stainless model to see if it does the right thing and just uses all four screws.

 

• The Meanwell power supply appears to be $28 by itself. 3+ Pole switches start to get considerably more expensive than single or dual pole equivalents, and theres no guarantee that there exists a 3-pole that physically fits and mates with the enclosure sliders. Relays are expensive as well. Now, at my low volumes, my solution will probably cost more than the combination of meanwell supply and hacked together or proto-board internal switch solution, but... It will truly look like it was made for the console, albeit looking more modern than anything else in the box. :D I may offer bare boards for those who want to order the parts from DigiKey and solder their own, I guess. I'd like to see what the final cost and demand are first.

 

• The Meanwell solution is still a "floor brick" albeit smaller and lighter than the original(see paragraph 2 of my original post). Moreover, you're locked into that supply or finding a similar supply with the same rails and 5-pin DIN power connector. My solution will allow a broad range of popular wall supplies, possibly many so compact that they plug in without obstructing other outlets on the wall or power strip as well as a range of voltage and current compatibility provided the total power is enough for the user's system(again needs will vary from user to user). All with a common 5mm(or 5.5mm I forget) OD, 2.1 or 2.5mm ID barrel type power connector at the back of the unit. I believe this provides the broadest appeal as many users can repurpose one of the may power adapters they already have laying around.

 

• Also, just a comment on MeanWell - I've seen some of their supplies. They appear to make cheaper "knockoffs" of supplies made by a more reputable company, CUI. However I've seen differences when comparing the two supplies as an exercise for a product team I worked on. They often use cheaper or smaller caps for the large bulk caps needed in such supplies, including caps with lower hour use lifetime ratings. Sometimes omit components that are not "strictly necessary", but improve the reliability or the quality of the power(ripple, droop, etc). I'm not a power supply expert, but I've had quite a bit of exposure to people who are and I've learned just enough from them to make good quality DC/DC switchers for my own projects and for projects I work on at my day job.

 

• Finally, this project is actually the core of three power supply projects which are not form factor compatible, but It would be desirable to share the individual switcher designs for once these have been tested/characterized. The TI-99 supply is first because it's the simplest of the three. I also plan a redesign of the internal switcher for the Apple IIc, so that it can be used with a smaller external adapter, preferably the same unit I'll use for my TI so I only need one packed in my vintage computer "trunk". The third is a full analog board/flat panel conversion for the Macintosh Color Classic. I'll probably need a beefier external adapter for that as the backlight for the TFT adds several watts to the total power as well as possibly needing to still support mechanical SCSI drives(although for my own unit, I'll probably also swap in an SDCard based replacement).

 

Again, not trying to knock anyone else's suggestions, but I believe the unit I'm designing will both have the broadest appeal for the many who just play with their vintage TI's plus some reasonable amount of additional capability for those who want to develop things that draw power from the side port(Maybe someone wants to redesign the CF7+ or Nano PEB to draw power from the console instead of needing the external 5V supply?).

Share this post


Link to post
Share on other sites
9 minutes ago, Omega-TI said:

 

A USB port on the TI could be used for many things, just off the top of my head...

 

1) If it was UNSWITCHED as dedicated close by power tap for the Pi ZERO on the TIPI/32K so we would not have to shut down power all the time.

2) To power for our cellphones as we are using them as a WiFi hub for our TIPI's to call into the BBS, check the weather, or doing anything at MYTI99.COM.

3) IF UNSWITCHED, as a convenient place to charge our cell phones at home.

4) For a nightlight to aim at the keyboard for those late night sessions.

5) The next big thing.TM

Yeah... I gotta say this all seems a bit too niche for the additional effort. Unswitched power for a trickle converter to turn on the main power is one thing, but unswitched charging power means there would be twice as many high power 5V regulators on board.

 

For switched power to an external Pi-Zero, I'll be providing extra capacity to the side port. I'm not sure what interface adapter is used for TIPI, but again, perhaps someone will do a design rev to allow the option to draw from the console(either with a jumper, or by "Diode-ORing" with low Vfwd Schottkys).

 

As far as the next big thing? I'm working on my version of that. Hint it's not an interface or peripheral designed for vintage computers, but I think it will appeal to a lot of vintage aficionados. That's all I'm gonna say about that at the moment. :P

 

Share this post


Link to post
Share on other sites

Yeah I guess the Mean Well would be more conducive to the unswitched aspect as it could be branched off before it goes to the TI as it has a dedicated 5v line.  For your design not so much as it would need more circuitry to step down the voltage.  Just tossing out ideas.

Share this post


Link to post
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

Loading...

  • Recently Browsing   0 members

    No registered users viewing this page.

×
×
  • Create New...