JSR03 going a bit nuts on me

[raq]

2.5, as the reference, both non-inverting and inverting inputs.
aR

[peranders]

The other pins?

If you have the same voltage (within mV's?) I'll suspect the opamp is working. The "sense" connections are tied to the output and ground?

[raq]

Yup, same voltages present.

While the op amp is indeed working now, except for the strange Vout, I suspect it is going to flip on me again. Let me give you an example. yesterday, after changing the op amp again (this time to a different type, from MAX410 to LME49710), I let it run with no load for 7-8 hours, just to make sure. It did great. When I sat it inside the enclosure, the blue LED would lit for a second, then turn off and I'd get 12V out (by the way, everything in the load should fry at 12V, I'm very lucky everything's still working).

Can you figure what's the reason for instability? Did I make some strange choice of parts? (I'm thinking things such as: maybe the op amp is too wide in bandwidth, and therefore it oscillates?... I don't know).

Can we maybe think of alternative scenarios?

Thank you,
Radu.

[raq]

yes, indeed, the 'current sense' outputs are jumped through solder underneath.
aR

[peranders]

Is H1 a red or blue LED? A blue LED is not recommended if you have set the LM317 to 2.5V. The blue LED has a higher voltage drop. I think you must tune up the LM317 to 3.5-4V. This means 6-8 V more in than than out.

Do you have 12 V out? What is the output voltage of the opamp?

(You could also use parts according to my recommendations.)

[raq]

Is H1 a red or blue LED?
>red

A blue LED is not recommended if you have set the LM317 to 2.5V. The blue LED has a higher voltage drop. I think you must tune up the LM317 to 3.5-4V. This means 6-8 V more in than than out.
>I have a blue LED as DZ1 - what you are saying above still applies?
If yes, how would I go about tuning the LM317 for higher voltage? Can I actually get rid of it altogether?

Do you have 12 V out?
>when things go bust, that is what I would usually measure at the output of the SuperReg.

What is the output voltage of the opamp?
>you mean, V at the output pin of the op amp? I will need to measure that tonight.

(You could also use parts according to my recommendations.)
>there is no BOM for 7.2V out, and besides, I am following the recommendations concerning each part. Most parts are taken from existing BOMs, and for the others I calculated everything by the given formulas, or tuned them by the given recommendations.

Do you want me to send you a scan of the schematic with all parts marked exactly as I have them?

Thanks,
Radu.

[peranders]

If you have a red LED in H1 position, that's fine.

Is the opamp in a socket? Are you sure you don't have a bad contact?

Can you check the opamp inputs when you are having 12 V out? If you have bad contact the inverting input (pin 2) should be more than 2.5 V and the output should be high. If you could take a photo so I can see the parts it would help.

[raq]

Hi P-A,

I just measured Vdc @ OUT (pin 6 of opamp), it is 1.787V, on average, it goes up and down ever slightly.

This is, of course, during stable operation. I will have to jolt it somehow, (maybe connect the DAC) to get it to flip and output 12V. I am currently having it run with no load, and during such operation, it is usually stable.

The op amp is indeed in a socket.  I do not know if a bad contact is at fault. The board is really small and a bad contact could go unnoticed unless I'd have a microscope to be 100% sure.

I will try to take some pictures and send over.

Thanks
R.

[peranders]

My guess is a broken transistor. Could you check the base-emitter voltage? Should be 0.65 V or around that. The output of the opamp should follow the output voltage with a DC shift only, LED+ Vbe.

[raq]

P-A,
The transistor you are talking about is the one at the output of the opamp, correct?
Thx,
Radu.

[peranders]

Yes.

Checked your pictures.

R10 should be 0 ohms and C9 not mounted. Those are "just in case".

T4, which type? Can't see.

Are the heatsinks isolated, not connected anywhere?

If the T6 is OK you could test to feed the opamp from R12 (Remove R11). MAX410 should work from 4.8 volts but you'll never know.

Nothing is connected to the output but to the sense inputs. You have tin blobs on the solderside in order to tie these inputs to the output and the ground?

[raq]

P-A,
Replies below.

Yes.

Checked your pictures.

R10 should be 0 ohms and C9 not mounted. Those are "just in case".

I think I will quickly build another one (have three more PCBs available). Will get this in there as indicated. Also, shall I keep R14 = 47.5ohm? Is sets 25mA CCS current.

T4, which type? Can't see.

D44H11.

Are the heatsinks isolated, not connected anywhere?

They are attached to the T's body, so I guess that's "suspended." No contact with anything.

If the T6 is OK

I hope it's OK, I just changed it when we thought the opamp was bad.

you could test to feed the opamp from R12 (Remove R11).

I am a bit more confident to do this with LME49710, as it can take more Vcc.

MAX410 should work from 4.8 volts but you'll never know.

Nothing is connected to the output but to the sense inputs. You have tin blobs on the solderside in order to tie these inputs to the output and the ground?

Affirmative.

Will keep you posted.
Thank you.

[raq]

testing... I made all of the changes, now it's working under load and I have my backlit fluke sitting there so I'd see the B+ at all times...

Will test it tonight until the end of the day (and a bit beyond, probably).

Will post here again, my morning.

Thank you,
Radu.

[raq]

Still good, P-A... I think we nailed it down.

Something tells me that feeding the opamp from the input instead of output made its operation more stable and fixed some delayed effects (I think there was something similar to a 'hysteresis' effect going on here, bouncing waves aggravated and amplified by the opamp loop). Rough startup and a positive feedback effect (making the voltage creep up slowly from the intended 7.2 to 12V out), due to the extra parts around the opamp, maybe.... This is sort of my understanding of what was happening.

It seems to be stable. I will keep you posted on this. A 5V and a 3.3V might follow (whoa!).

Thanks for your assistance, appreciated,
Radu.

[raq]

P-A,

Whenever you get a chance - further comment on all this...

As I am running about 25mA on the CCS, I thought I should monitor the temperature of T1/T6. T1 is in a very safe region, of a little over 30C. T6, though, heats up to just over 60C.

Would you regard that as safe? I am not sure what temp as should regard as bulletproof in that position T.

Please let me know.

(By the way - I've been monitoring the voltage - no problems this far).

Thank you,
Radu.