FAQ & Terms of UseFAQ & Terms Of Use   Wiggler RadioMW Radio   Muff Wiggler TwitterTwitter   Support the site @ PatreonPatreon 
 SearchSearch   RegisterSign up   Log inLog in 

RPSU for 1Amp +-15v
MUFF WIGGLER Forum Index -> Oakley Sound Systems  
Author RPSU for 1Amp +-15v

Have been trying to load the RPSU with 1Amp delivery, but after a few seconds it fails and the output voltage of both pos and neg rails is close to zero.

As the regulators are capable delivering that load, was wondering whether I need to change some of the caps from the original spec.

Which once should I upgrade and with what?

The regulators are rated to deliver 1.5A but that is their maximum. However, they will only do so if their temperature is kept below a certain level. They have an automatic thermal shut down if their die temperature exceeds a certain level.

This means that they will only deliver 1A if they are attached to a decent heatsink. Having no heatsink will make them switch off pretty damn quick.

The speed at which any device heats up is related to not only the current but the voltage drop across the device. So even if a decently mounted heatsinked LM317 will be able to cope with a 1A output with a 18V input it will perhaps overheat with a 25V input voltage.

Thanks Tony.

I have connected the board and trannies to an aluminium back panel, so heatsink is relatively large (2 module high, 19" analog rack). Current draw is around 0.7A.

I measured LM317/TL337 input at it is at 26v, which means it might be heating up too much (am using a 18-0-18 transformer).

Is there a way to reduce voltage going into the LM317/337 more towards 18v or so?

Alternatively, does using an LM350 (3A) alleviate this issue?
Perhaps using a TO-3 version will allow for higher amps?
I think you may have done something wrong with the build. Once properly heatsinked the devices should be fine with that sort of current.

What are the voltages like without any load? That is, just the power supply connected up but no modules attached.

With no load, each transistor sees about 26v at its input.
With load, it is about 24v.

I have connected a LED to the output 15v just to see when things get bad and after 10-15 sec of full load (0.7A), voltage on both rails drop to almost zero and the LEDs are fluctuating at a rapid rate, like 10 Hz or so.
I'm not sure, but it does sound like too much current is being taken by the load. How sure are you on the 0.7A figure? Have you measured it with meter? It might be worth powering it up with a ammeter wired between the +15V output of the power supply and the +15V connection of the load.

I take it the RPSU produces a stable +/-15V without any load connected?

Yup, have measured it with a fluk on the positive rail between the RPSU and the power distribution board. Measure is 0.7A.

Is there a chance that a component prior to the transistor is under spec for this load?
Well, I'm stumped. It does sound like a thermal shut down issue though. How are you mounting the device to the heatsink?

whomper wrote:
Is there a chance that a component prior to the transistor is under spec for this load?

If you have used the devices given in the Builder's Guide there shouldn't be a problem. But if there is a problem it would show up as a voltage drop on the input to the regulators, and/or that some of the devices would get very hot.

It is interesting that both regulators shut down at the same time. This does strike me as very odd. Even with the same current going through each, they are different devices and have slightly different specifications.

What is supplying the power to the RPSU? Are you using full wave rectification or just half wave?

One other question: What have you got for the fuses?

The devices mount in a similar way to suggested image. Have attached an image to demonstrate.

I guess both shutdown at the same time as they are connected to the same heatsink and share the same temp.

Am using full wave with a centre tapped transformer.

Fuses are standard ones, not slow blow or anything AFAIK.

Will be connecting a 5A TO-3 as the positive regulator to see what it will do.
I can't tell from the photograph, but did you use any thermal grease between the transistors and the panel?

If you haven't, you may have a bit of a thermal break between the transistors and the panel that is decreasing the thermal conductivity. For the panels I build for the Oakley standard PSU, I used some thermal grease on both sides of a thin rubber insulator to thermally connect the transistors to the Aluminium panel. This is especially important with some anodised panels, as the anodising can make the surface quite rough, limiting the thermal contact.


A few thoughts:

That looks like a TO3 mica washer under U2. As Magman says mica really needs a bit silicon grease to work properly. It's hard to tell what's under U1, but again, if its mica, it'll need some grease.

Secondly, the panel itself looks as if it's not too thick. A thin panel will struggle taking the heat away from the devices. And if it's steel rather than aluminium it'll be even worse. I'd recommend 2mm aluminium at the least.

Thirdly, with a 18V winding you'll be getting around 25V or more at the input to the regulator. That's 7W of power to be dissipated in the device which is quite a lot to get rid of. It may be better using a 15V transformer instead.

Thanks Magman and Tony,

Have added thermal paste to both transistors as well as added another transistor set at the top with their own heat sink, soldered on top of the first transistor pair. This should split the generated load between the two transistors per rail, reducing overall heat.

So far, so good. Heat is generated at the four transistors, but does not reach meltdown levels. Have been running it for over an hour and it is stable.

Hope this puts this item behind me :-) Onwards to complete an HVM/SE330 combo!
Hang on, did I read that right? Are you simply connecting one regulator in parallel with the other? If so, you shouldn't do this. One regulator will be doing most of the work while the other does little. It may also be unstable and prone to oscillation.

You can use current sharing resistors (see National's data sheet) but this is not necessary for currents less than what a single 317 is capable of. I think you should look at properly heatsinking the single devices. The general rule of thumb is that if you can't touch a device for more than ten seconds it's too hot.

Good thought on the current sharing resistor. I'll have a look at that to see how I can leverage while splitting the load between the double transistors.
MUFF WIGGLER Forum Index -> Oakley Sound Systems  
Page 1 of 1
Powered by phpBB © phpBB Group