Pumpkin Problems

[AndrewT]

Replace the IN to OUT diode with a 30V Zener.

So you were talking about your Zener then, not the one in my circuit.

Yes.

 

[AndrewT]

Is there any possibility the Nch and Pch FETs are swapped?

 

[KatieandDad]

I was confused.

I knew that the 10V zener was perfectly safe.

I'm on the point of giving up with the Pumpkins anyway as I cant stop them oscillating. Just going to try the other one to see if its the same.

GRRRRRRRRRRR

 

[KatieandDad]

In (+) In (0V) and in (-)

Are all shorted on the board.

There is nothing else connected except for the PSU.

 

[KatieandDad]

Both Pumpkins are exactly the same which almost rules out component failure.

The odd thing is that the regulators are providing a clean +/- 36V into resistive loads.

As soon as a Pumpkin is connected, the 20MHz waveform appears on the supply rails, even when bypassed with 220uF + 220uF + 10uF + 120nF + 220nF.

 

[KatieandDad]

Funny thing is they "work", they just have this strange "whooshing" noise.

 

[KatieandDad]

All components are the correct value and the correct way round.

I'm at a complete loss as to why they are oscillating.

I even suspected the 'scope but another shows the same story.

 

[KatieandDad]

Maybe the excitement of the Pumpkin is this oscillation. Andrew T came up with a different approach saying, maybe they are supposed to oscillate.

 

[KatieandDad]

Funny how all the "Design Engineers" out there cannot even offer an explanation as to why the Pumpkins are oscillating.

I have a feeling that they all do this and that I am the first to notice it.

 

[Salas]

Both Pumpkins are exactly the same which almost rules out component failure.

The odd thing is that the regulators are providing a clean +/- 36V into resistive loads.

As soon as a Pumpkin is connected, the 20MHz waveform appears on the supply rails, even when bypassed with 220uF + 220uF + 10uF + 120nF + 220nF.

I would delete one by one the decoupling capacitors hanging off the supply rails on the Pumpkin starting with the 120n & 220n. Checking if the oscillation would back off or even stop. There could be a chance they twist the phase margin of the supplies.

 

[Bonsai]

You have some solid decoupling on the rails. If the PSU runs ok with just a resistive load instead of the Pumpkin, you need to look at the Pumpkin, not the PSU.

Has this circuit been simulated to look at the loop response? I have taken a cursory look and it seems the OLG is quite high (if I am reading the circuit correctly!).

There is no compensation, and with th lead cap in the feedback resistor, matters are only made worse. The other option here is to increase the LTP degeneration resistor- currently at 47 Ohms. Try replacing this with a 500 Ohm pot. If that stops it, it's probably your compensation design that needs a bit of work.

 

[KatieandDad]

I'll give it a go tomorrow. I wonder how many people are happy with this poor design.

 

[Bonsai]

The open loop gain of this designs about 30 000 assuming a very light output load. You definitely need to compensate it. Try the pot change I mentioned above, and let us know how it affects the oscillation. After this we can take the next step with the compensation design. I am sure this thing can be fixed.

 

[BDP]

The open loop gain of this designs about 30 000 assuming a very light output load. You definitely need to compensate it. Try the pot change I mentioned above, and let us know how it affects the oscillation. After this we can take the next step with the compensation design. I am sure this thing can be fixed.

Could you explain where you get the 30,000 OL gain figure?

BDP

 

[Bonsai]

My Mathis was not too good after 3 glasses of red. It's about 4500 and not 30 000.

Input stage is about 15x and output stage is about 330.

It still needs some serious comp to make it reliably stable IMHO.

 

[BDP]

Thanks for the reply.

That is what I figured. I wonder what the optimum gain figure would be. I think Nelson said that Super Symmetric feedback worked best with simple linear circuits, does that mean lower gain figures? It would be easy to load the output to reduce the gain. Or use degeneration on the sources of Q1 and Q2.

BDP

 

[tonimxp]

Pumpkin problem Upgrade!

Trying also the craziest things to solve the problem, I finally solved!

I removed R21 10k and it all worked out!

Then i decided to investigate and i found out that there is also a version of Pumpkin using in place of R21 ... U1 LM336Z-5V

Pumpkin Preamp - Perfect for F4

Question... What should I do now leave everything just as it is now, or change something?

Removing R21 can involve some variation on the frequency response / stability of the circuit, or ......

I feel comfortable, listen to music and drink wine?

I hope that this discovery could also help other people in this forum!

Greetings to all.

 

[tonimxp]

Pumpkin noise problem

Hi Bonsai! I noticed that there are several differences between the schematic 1
( that reflects the reality of my pcb) and schematic 2 .
I noticed that i reconnecting R21 but bringing R22 from 47 to 220 ohm things
got better, then I also connected to the base of Q9 between D4 and D5.
But to be able to adjust the offset between -out and +out i had increase
WR1 from 47 to 220 ohm.
Now Noise is OK! (please see shematic 3)
Strangely I have a gain of 4 instead of 4.7 using R5 and R6 47K and leaving R1
and R3 10K, I also changed C1 and C2 from 5p to 10p.
I think everything I've done.
Now i need a greater extent in the high frequencies, I realized that my Pumki
vs my Emerson is less extensive on high.
Please can you give me some advice for more extension?
I believe that pumpkin can give more.
Thanks Bonsai

Be well!