Are you sure that the xformer can handle the DC current from the DAC? Low level transformers often distort even at the smallest DC current.
I bet that when you remove the xformer and insert a signal at point A it will look fine and dandy. If that is the case, the cause is identified and the rest is just straight forward engineering
Jan
I bet that when you remove the xformer and insert a signal at point A it will look fine and dandy. If that is the case, the cause is identified and the rest is just straight forward engineering
Jan
Last edited:
The OP275 is a butler input stage based opamp, since nothing so far has resulted in one iota of difference in the issue I recommend trying a conventional fet opamp like the OPA2132 or 2134.
OPA2132 | Precision Op Amps | Operational Amplifiers (Op Amps) | Description & parametrics
OPA2132 | Precision Op Amps | Operational Amplifiers (Op Amps) | Description & parametrics
That's a good point. That's why I checked and double checked that DC is as low as possible on the primary of the xformer. My DVM says 0.000 V. Doede (the guy who designed the DDDAC) tested my xformer in his own DDDAC, which has a small DC imbalance of about 5 mV. He found no issues with his AP distortion tester (see attachment).
That's an easy bet as I already described this in my first post
"I replaced the DDDAC/TVC part with a signal generator (50 Ohm output). The distortion went away."
Uhm, really? Remember, the signal coming from the xformer is clean (at point A). The distortion shows only at points B and C. I'm not sure I understand this.
Attachments
Ok, as far as I can tell, I've got two more to do items on my list:
I will be quite busy with other things in the next few days, so I might be a bit slow with this. But I will make it somehow...
I will be quite busy with other things in the next few days, so I might be a bit slow with this. But I will make it somehow...
Let's try to recollect what you have done to prevent turning in circles.
1) when driving your filter with the trafo, you have a clean signal at point A, but not on B & C.
2) when driving with a 50 Ohm signal generator you have no distortion [I missed this point].
3) when removing the 100nF filter caps while steering from the trafo, distortion at C is still there.
4) when inserting a buffer between the trafo and point A of the filter, distortion at A,B and C is as in point 1).
Is this a correct presentation of facts ?
Especially 2) and 4) seem to be in contradiction since one may assume the buffer has a very low output impedance.
The op275 might arguably be not the best opamp for this function, but it should without a doubt be able to perform properly.
I still have the feeling something could be wrong with:
A) the signal ground, or
B) Your op275 is defective because of accidently having the output short circuited with one of the supply lines while measuring,or
C) You have a fake op275 from China.
In case the op275 is socketed, you could replace it with almost any other opamp with the same pinning, just to see if it makes a difference.
Hans
1) when driving your filter with the trafo, you have a clean signal at point A, but not on B & C.
2) when driving with a 50 Ohm signal generator you have no distortion [I missed this point].
3) when removing the 100nF filter caps while steering from the trafo, distortion at C is still there.
4) when inserting a buffer between the trafo and point A of the filter, distortion at A,B and C is as in point 1).
Is this a correct presentation of facts ?
Especially 2) and 4) seem to be in contradiction since one may assume the buffer has a very low output impedance.
The op275 might arguably be not the best opamp for this function, but it should without a doubt be able to perform properly.
I still have the feeling something could be wrong with:
A) the signal ground, or
B) Your op275 is defective because of accidently having the output short circuited with one of the supply lines while measuring,or
C) You have a fake op275 from China.
In case the op275 is socketed, you could replace it with almost any other opamp with the same pinning, just to see if it makes a difference.
Hans
Excellent question! We need to gather data first before randomly suggesting anything as a cure
. Me included.Jan
Yes.
Yes. It's like the filter opamp can smell the transformer through the buffer.
Yes, it's socketed. I already tried an NE5532 a few days ago, and now also an LM4562. The distortion effect was the same (the signal shape of the distortion may have been a wee bit different, but not much). As far as I can tell, these opamps all have bipolar inputs. So the FET input idea might indeed be something to test. But I don't think my junk box has anything to offer here, so I'll have to get some first. While I'm at it I might just as well get a complete Sallen Key filter module with FET input opamps (there are some at eBay), just to make sure the culprit is not in the PCB design.
Excellent question! We need to gather data first before randomly suggesting anything as a cure
Jan
Take a look at my first post and also post 25.
In summary:
- The distortion is most apparent at the lowest volume setting of the TVC and gets less apparent at higher volume settings.
- I set the TVC to "12' clock volume" and attenuated the test signal in the software before it was sent to the DAC to get the same signal amplitude at the TVC output. The result is that the distortion at the filter output was gone.
This smells like parasitic oscillation. What happens if you insert a pi or T attenuation pad between the TVC and the filter? What happens if you insert a Zobel network at the input?
My guess is that the opamp, although unity gain stable, presents a negative resistance at some HF frequency and this happens to react with the impedance presented by the TVC.
My guess is that the opamp, although unity gain stable, presents a negative resistance at some HF frequency and this happens to react with the impedance presented by the TVC.
Almost surely it is an oscillation, as other people already indicated, but no relavant answer or act had happened.
There are many possible reasons, and not neccessarily restricted to the OPA in the filter, but for example the unused OPAs also can oscillate if not deactivated correctly.
The root cause of distortion is
- NOT the transformer, since A is undistorted
- not the OPA alone
- affected by leakage inductance
A high speed oscilloscope is needed to check oscillation.
Probably the leakage inductance with input and other parasitic capacitance forms a parallel resonant circuit near the GBW product of OPA where the fully fed back OPA has gain, and can maintain oscillation.
I have simulated the filter circuit in LTSpice and no matter how I try to screw up the circuit, I can't get the waveforms that you show us.
It is as if a spike of very short duration is overloading the input of the filter, leading to this effect.
That could explain why your input signal seems clean, because your scope is not fast enough.
That would also explain why a buffer does not make a difference because it simply transfers the spike.
Jan's suggestion to try other frequencies could possibly confirm whether the distortion is directly related to the input frequency.
Does your scope have the ability to catch spikes ?
Hans
It is as if a spike of very short duration is overloading the input of the filter, leading to this effect.
That could explain why your input signal seems clean, because your scope is not fast enough.
That would also explain why a buffer does not make a difference because it simply transfers the spike.
Jan's suggestion to try other frequencies could possibly confirm whether the distortion is directly related to the input frequency.
Does your scope have the ability to catch spikes ?
Hans
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.