exxos's Atari, Amiga & retro forum

rubber_jonnie wrote: 20 Sep 2023 15:42 What issue is that Stephen? Mine is closer to the issue 5 (I can't find a 4s specific schematic) IIRC I did check for shorts in the coil but I can do so again. It's a fairly rudimentary device that's for sure.

Well, as with any transformer, its the ratio of turns in the primary relative to the secondary which determines the voltage change. So if a few of the turns are bypassed by a short then the ratio will change and the voltage will change.

Voltage on pulse transformers are more dependent on the on off switching. Longer on time means more power dumped into the core where voltage will rise if the energy exceeds the current requirements by the load.

It's why switch modes can explode without a load as the pulse can cause a voltage rise over 50+ volts. It's why it's important to have ovp etc on them.

For arguments sake, if the secondary had 50 turns and 10 were shorted. The feedback loop would compensate by dumping more energy into the core. Mostly will run hotter and less efficient.

So maybe there's no-longer enough load on the +12V and -5V lines now that the RAM is no-longer taking that energy?

stephen_usher wrote: 20 Sep 2023 16:31 So maybe there's no-longer enough load on the +12V and -5V lines now that the RAM is no-longer taking that energy?

Looks like it's a step up from 9v to 12v. So if no load on the 12v the regulation wouldn't work and could well spike up to 50v. Ive not been following close enough to know the test done.. But if no load then the behaviour would actually be normal.

exxos wrote: 20 Sep 2023 16:35

stephen_usher wrote: 20 Sep 2023 16:31 So maybe there's no-longer enough load on the +12V and -5V lines now that the RAM is no-longer taking that energy?

Looks like it's a step up from 9v to 12v. So if no load on the 12v the regulation wouldn't work and could well spike up to 50v. Ive not been following close enough to know the test done.. But if no load then the behaviour would actually be normal.

There's some load, the video encoder chip, but not as much as if the 4116s were in circuit. The -5V and -12V would also be fully unloaded.

@rubber_jonnie You could replace the whole of the spectrum DC-DC conversion circuit with one of these: https://www.ebay.co.uk/itm/265904298733

stephen_usher wrote: 20 Sep 2023 18:07

exxos wrote: 20 Sep 2023 16:35

Looks like it's a step up from 9v to 12v. So if no load on the 12v the regulation wouldn't work and could well spike up to 50v. Ive not been following close enough to know the test done.. But if no load then the behaviour would actually be normal.

There's some load, the video encoder chip, but not as much as if the 4116s were in circuit. The -5V and -12V would also be fully unloaded.

@rubber_jonnie You could replace the whole of the spectrum DC-DC conversion circuit with one of these: https://www.ebay.co.uk/itm/265904298733

Certainly cheaper than buying a new coil and TRACO

stephen_usher wrote: 20 Sep 2023 18:07

exxos wrote: 20 Sep 2023 16:35

Looks like it's a step up from 9v to 12v. So if no load on the 12v the regulation wouldn't work and could well spike up to 50v. Ive not been following close enough to know the test done.. But if no load then the behaviour would actually be normal.

There's some load, the video encoder chip, but not as much as if the 4116s were in circuit. The -5V and -12V would also be fully unloaded.

@rubber_jonnie You could replace the whole of the spectrum DC-DC conversion circuit with one of these: https://www.ebay.co.uk/itm/265904298733

Sorry I didn't reply until this morning, I was in the office yesterday, and it's a very long day for me.

I tested the coils on both the + and the spares board again, there are no shorts between windings.

So I got some new TR4s and fitted one to the good board. 30v this time but I removed power PDQ and had already pulled the ULA and LM1889 for safety, so no harm done on this occasion. I pulled the transistor to re-use on the spares board.

When installed on the spares board, which still has the 4116 RAM installed, the results were different, I got a steady 7-8v for about a minute and then the transistor popped. It seems like these 4s boards are unhappy fellows and like to pop TR4s like no tomorrow. The RAM may have dead IC's though, causing an overload condition that pops TR4.

So, I'm going to pull the 4116 RAM from the spares board next and fit another TR4 and see if the results match that from the + board, i.e. 20-30v without the load, or if it runs happily and delivers 12v.

But I'm guessing without the load on -12v and -5v it's going probably going to run away again.

I could remove the hacked 4164s and drop in the spare 4116's I have to determine if the load changes with the -5v & -12v rails in play, however if I get 20-30v again, that's my spare 4116s lightly toasted. I'm not keen on that option.

So I think it likely that I need to do 1 of 2 things.

The PSU provides 11v, and if I steal that from the input, I can inject that into the dead 12v circuit (At D15, on the same side as C78) and the machine runs just fine. I'm not overly keen on that idea since it's taken directly from the PSU input and unregulated. If the PSU goes bad, the consequences could be catastrophic. Though given the 30v I have been getting on the original circuit, is it that much worse?

Alternatively, there is the suggested boost regulator. I can feed it 5v from the regulated power line, and it should offer a level of isolation from the PSU. Time to get ordering.

For reference the issue 5 circuit is below:


EDIT: Boost converters ordered.

If it is a load issue then you could put a 50 ohm resistor on the +12V and -5V lines to ground to add a reasonable load even without the chips.

Also sounds like your transistors are too low current versions.

stephen_usher wrote: 21 Sep 2023 10:48 If it is a load issue then you could put a 50 ohm resistor on the +12V and -5V lines to ground to add a reasonable load even without the chips.

Also sounds like your transistors are too low current versions.

The transistors I have now are as per the schematic, so they should be fine, though they could have been underspecced from the outset.. I'm done blowing up transistors now though, just going to use an alternative to the 12v section.

I have boost regulators on the way now, so that will be the fix, though I have now removed the 4116 RAM from the spares board and see what it does now without load, and I can see if I can sort out some sort of artificial load on the spares board too, though not sure if I have anything suitable to hand.

I just think the boost converter will be a lot more stable than this thing.

Ok, so that was interesting.

I removed all the 4116 RAM from the spares board and fitted TR4 back in. Yet again, a 50% fail rate when testing the 4116's, and I have to say some of the previous rework on this board is pretty awful, as in lifted tracks, burned on flux etc, so I'm giving it a good cleanup as I go.

Power on and I have -5v, +5v & 12v all present and correct. I wasn't expecting that, so I guess my original issue 4s board has some sort of issue, though I'm not sure where. I'm guessing it isn't load related at this point, but some other component failure in the system.

It looks very much like the old faulty RAM was the cause of the blowing transistor.

So next I'll install some sockets on the spares board, since with a ULA fitted and no RAM I'm getting a video signal, so it seems this board may be a runner.