Help needed!!
In the process of building my Pearl II, and I encountered a difficult problem . One channel adjusts just fine, down to a few tens of mV offset, the other channel doesnt.
When I turn the pot, the offset voltage will come down from around 12V down to maybe 3V, then suddenly jump to -22V. When I turn back up it will jump from -20V to around +7V or higher.
Checked all the passive components and the wiring, soldering points etc, replaced all active components at least three times, measured all impedances at all nodes and compared to the working channel, verified all the voltages with the working channel and the schematic posted with the voltage measurements multiple times.... and the problem persists.
When first powered up, both channels were happily oscillating at around 12MHz, and so I increased C7 and C9 to 24pF, and added 100pF across drain-source of Q2, and another 33pF across R14 (most counter-intuitive, but it worked) and all the oscillation stopped (leave one of these out, and it restarts, at low level but it will oscillate...). The only other modifications are: ZVP3306 instead of ZVP3310 for Q2, and 1uF foil cap at the output instead of a 22uF electrolytic. All of these changes should not affect DC balance, though.
Stuck at an impasse here.... any help highly appreciated!!!
In the process of building my Pearl II, and I encountered a difficult problem . One channel adjusts just fine, down to a few tens of mV offset, the other channel doesnt.
When I turn the pot, the offset voltage will come down from around 12V down to maybe 3V, then suddenly jump to -22V. When I turn back up it will jump from -20V to around +7V or higher.
Checked all the passive components and the wiring, soldering points etc, replaced all active components at least three times, measured all impedances at all nodes and compared to the working channel, verified all the voltages with the working channel and the schematic posted with the voltage measurements multiple times.... and the problem persists.
When first powered up, both channels were happily oscillating at around 12MHz, and so I increased C7 and C9 to 24pF, and added 100pF across drain-source of Q2, and another 33pF across R14 (most counter-intuitive, but it worked) and all the oscillation stopped (leave one of these out, and it restarts, at low level but it will oscillate...). The only other modifications are: ZVP3306 instead of ZVP3310 for Q2, and 1uF foil cap at the output instead of a 22uF electrolytic. All of these changes should not affect DC balance, though.
Stuck at an impasse here.... any help highly appreciated!!!
This is mosfet dependant maybe but also layout dependant. Which one do you currently use ?
For the offset thing : have you tried swapping the trimpot from the faulty board to the other one ?
The 2nd stage has roughly 80 dB O.L. versus 40 or so closed loop original values; 40 dB roughly equals x 100 times. You get a kind of equilibrium between closed loop gain / feedback.
Getting a 33 k resistor leads you to diminuish the closed loop of the 2nd stage, approx. 33 times so roughly around 30 dB. So it means you input about 50 dB feedback. This can lead the design to be more prone to oscilations problems in some cases. The method you described to get rid of them is fine and component / desired gain / layout dependent. For instance, at unity gain maybe the 2nd stage would need even more compensation than the one you applied
Regards,
nAr
Thanks nar for your answers, currently using the ZVP3306.
On the compensation, yes you are right, with a lower impedance there the feedback cap would have to be tripled (30pf). I will try and measure the OL bandwidth once the thing works.....
On the trimpot, I just replaced it with a brand new one, and measured it before putting it in, as well as the 1.5k resistor and their impedances to ground etc. everything was correct, still, the problem persists even with the new trimpot. the other channel "just laughs in my face" and continues to operate just fine.... sigh.....
Another thing I thought was that the two ground planes are connected in one point (via at the center of the PCB), I measured both and found some dubious values for the non-working channel (7 ohms), as compared to zero for the working channel. put a nice piece of wire across that connection (on both boards), measured both to have zero resistance there, and guess what..... the problem still persists.....
having a nice glass of wine now, ignoring the phono preamp and listening to a CD ;-(
On the compensation, yes you are right, with a lower impedance there the feedback cap would have to be tripled (30pf). I will try and measure the OL bandwidth once the thing works.....
On the trimpot, I just replaced it with a brand new one, and measured it before putting it in, as well as the 1.5k resistor and their impedances to ground etc. everything was correct, still, the problem persists even with the new trimpot. the other channel "just laughs in my face" and continues to operate just fine.... sigh.....
Another thing I thought was that the two ground planes are connected in one point (via at the center of the PCB), I measured both and found some dubious values for the non-working channel (7 ohms), as compared to zero for the working channel. put a nice piece of wire across that connection (on both boards), measured both to have zero resistance there, and guess what..... the problem still persists.....
having a nice glass of wine now, ignoring the phono preamp and listening to a CD ;-(
As an experiment try R16 at 100K. Also on one version I built, C7 caused oscillation. Remember it isn't fixed until you can break it and fix it at will.
The DC flip flop can be oscillation caused.
If I recall correctly Wayne suggested to try to leave out C7 at first. That's what I did and never ran into oscilation problems, DC was always stable and trimmable, there is a small drift but hey - nothing alarming, still around zero, and as I chose K170 instead of K370, as the plastic case is bigger, the offset wanders maybe a little bit more than if I would have used K370 ...
Hesener, which is the IDSS of the devices you chose for the diff. pair ?
Seems if the DC is not precisely trimmable; try to see if the DC polarization given on the ZVP by the ref. of the ajustable stays still. If so, you could say the ZVP has a problem around itself
But, if the DC reading on the trimmer - sorry, whole resistor array - wanders as the output offset, then your problem might be CCS based, e.g. in the bottom of the circuit. Try to see if all readings around the red led are correct, and do you have stable current flowing into diff. pair and output section ?ZVP3306 has some little different specs on the datasheet. Fine tuning around the device is indeed needed.
Hesener, can you post pics on some square waves at different frequencies that you would feed at the input of the K370 differential, and hooking the probe to the output, trying to get 1V amplitude at the output ( so according to gain, imput signal has to be small enough ) and could you try the procedure on both channels ?
Lastly, which layout did you use ? I mean, for the boards.
Don't have an oscilo to check any oscilation on mine, though. Sonically sound is clear, undistorded and fine detail layering, DC params are also stable so ...
Trying to help,
nAr
Algar_emi has done so here, for operating points values, may I show the post:http://www.diyaudio.com/forums/pass-labs/156397-pearl-two-63.html#post2448708 ( 2nd thumbnail )
Approx. gain values for feeback resistor values ( 2nd stage )
100k about 40dB (original)
50k about 33dB
33k about 30,4dB
150k about 43,5dB
Capacitor in parallel with feeback resistor will logically need to be increased for lower gain settings than the 100k original value
nAr
Last edited:
its working now....
thanks wayne and nar for your help, its working now!
I increased the feedback resistor back to 100k and removed C7, and it worked just fine on both channels - no oscillation, offset fine. I also removed the cap from the gate of Q5 to ground, no oscillation. very strange..... another (quite experienced) colleague and myself have never seen such a flip-flop behaviour. the only explanation we could come up with is that the gain must be very low once the circuit sits in one of the corners, hence it cannot charge that cap anymore. but that is just a guess, some simulation might reveal it better.
A quick check with a function generator made me add a 2.7k for Rx in one of the channels, now there is less than 0.3% of voltage difference between 100Hz and 10kHz. (Well, I did match all the caps and resistors to 0.1%, which took a long time and a big bag of caps....). This should be better than cartridge mismatch, tonearm adjustment or even the record itself.
Looking forward to the listening test tonight.... will post my impressions and a couple pictures then. Again, thanks a lot!
thanks wayne and nar for your help, its working now!
I increased the feedback resistor back to 100k and removed C7, and it worked just fine on both channels - no oscillation, offset fine. I also removed the cap from the gate of Q5 to ground, no oscillation. very strange..... another (quite experienced) colleague and myself have never seen such a flip-flop behaviour. the only explanation we could come up with is that the gain must be very low once the circuit sits in one of the corners, hence it cannot charge that cap anymore. but that is just a guess, some simulation might reveal it better.
A quick check with a function generator made me add a 2.7k for Rx in one of the channels, now there is less than 0.3% of voltage difference between 100Hz and 10kHz. (Well, I did match all the caps and resistors to 0.1%, which took a long time and a big bag of caps....). This should be better than cartridge mismatch, tonearm adjustment or even the record itself.
Looking forward to the listening test tonight.... will post my impressions and a couple pictures then. Again, thanks a lot!
The gain is set by the ratio of resistors only. 2nd stage is non-inverting. Gain is (R16+R14)/R14 or (100K+1K)/1K, 101 times or 20 log(101) = 40,08 dB. Resistors do have their own tolerance and ppm, and that added to soldering hazards and traces can lead to slightly different values once built. You did fine to get adjusted voltage gains in your particular configuration.
Glad you succeeded. You did it
This is also great to know people can use ZVP3306 instead of ZVP3310 without so much mods elsewhere in the circuit.Best,
nAr
big grin on the face
Pearl II is up and running, and I am enjoying myself. Feels like it is holding back a little bit still, but with more warm-up and more hours of burnin this will go away, I am sure. Noise is way down, only if I crank up my linestage (which has too much gain anyway) it is audible.
It does sound pretty good right out of the box, I like the clarity and also the tonal balance a lot! the soundstage is very solid, it has a good rhythm and pace.... Now I guess I need to get going on my tonearm adjustments ;-)
Thanks a lot Wayne for this nice design, and your help! Sounds great! And thanks to Nar for the help in getting it up and running!
best regards - must go back to listening mode....
Pearl II is up and running, and I am enjoying myself. Feels like it is holding back a little bit still, but with more warm-up and more hours of burnin this will go away, I am sure. Noise is way down, only if I crank up my linestage (which has too much gain anyway) it is audible.
It does sound pretty good right out of the box, I like the clarity and also the tonal balance a lot! the soundstage is very solid, it has a good rhythm and pace.... Now I guess I need to get going on my tonearm adjustments ;-)
Thanks a lot Wayne for this nice design, and your help! Sounds great! And thanks to Nar for the help in getting it up and running!
best regards - must go back to listening mode....
Be ready to try different grounding schemes to get least background noise
On my setup, PSU is deported into anther housing and couple of meters away. Then the Pearl II chassis is directly linked to Earth. Audio GND is separated from chassis. From input RCA, simple wiring ( twisted single stranded wire ) signal, ground, x2 ( hence 2 channels ). The 2 boards have V+ GND V- from single PSU. Then, from outputs of the Pearl boards, only + ( x2) to the output RCAs; GND is taken from psu entry into Pearl box and one wire splits to the 2 gnds of the output RCAs.
Last, audio ground is linked to earth via a thermistor at PSU level.
It's roughly the same principle as there was for the ZLS
Enjoy Pearl II
Best,
nAr
Cool, I did exactly the same thing. To get the grounding right is so crucial on these!
My power supply box is 1.5m away, with the grounds connected via a bridge rectifier. the "true" ground is a star ground at the input of the phono stage, and all wires go there. Both PCBs sit in copper chambers, that have a direct and individual connection to that star ground. Another advantage would be of course that I can use that power supply for trying other amplifiers. (I put pictures in the "pictures" thread, that may explain it better)
I am running a cable with 7 wires from the PSU to the phono stage, separating the voltages at the supply box already , using cap multipliers. These wires measure about 1mV of noise where they enter the box, then there is another noise filter after which I measure 0.5mV directly on the power supply connections of the Pearl II board.
So, when I turn up the volume I hear no hum, just a little bit of hiss (which comes from the phono stage - when I do that with any other input there is no noise). Running the mains cable of the CD player parallel and in close proximity to the phono in put cable does cause hum to appear, though (not surprising).
Next on the list is new cabling for the turntable, with short wire runs directly to where the input sockets on the phono stage are, to minimize the loop and also the capacitance.
best regards
My power supply box is 1.5m away, with the grounds connected via a bridge rectifier. the "true" ground is a star ground at the input of the phono stage, and all wires go there. Both PCBs sit in copper chambers, that have a direct and individual connection to that star ground. Another advantage would be of course that I can use that power supply for trying other amplifiers. (I put pictures in the "pictures" thread, that may explain it better)
I am running a cable with 7 wires from the PSU to the phono stage, separating the voltages at the supply box already , using cap multipliers. These wires measure about 1mV of noise where they enter the box, then there is another noise filter after which I measure 0.5mV directly on the power supply connections of the Pearl II board.
So, when I turn up the volume I hear no hum, just a little bit of hiss (which comes from the phono stage - when I do that with any other input there is no noise). Running the mains cable of the CD player parallel and in close proximity to the phono in put cable does cause hum to appear, though (not surprising).
Next on the list is new cabling for the turntable, with short wire runs directly to where the input sockets on the phono stage are, to minimize the loop and also the capacitance.
best regards
OK... I just completed my build and I'm getting some nonsensical results. Without
connecting the PS to the PII boards, I get +34VDC and -34VDC rails. When I connect
it to the PII boards, it changes to +24VDC and -44VDC... what's going on?
I have checked once for solder joints, and such and will do it again later today...
Meanwhile, if anyone has any clue as to what's going on, it would be great.
I have not adjusted the pot for zero dc offset yet, but that shouldn't affect my rails
correct?
connecting the PS to the PII boards, I get +34VDC and -34VDC rails. When I connect
it to the PII boards, it changes to +24VDC and -44VDC... what's going on?
I have checked once for solder joints, and such and will do it again later today...
Meanwhile, if anyone has any clue as to what's going on, it would be great.
I have not adjusted the pot for zero dc offset yet, but that shouldn't affect my rails
correct?
The GND point between the 2 bridges is not link to PSU GND maybe
?Remember there is an error ont eh Pearl II PSU schematic; the common point between the 2 bridges should be also the central GND point. An horizontal line is missing in the drawing
So can you show us your PSU, schematic and wiring ?
Correct. Pot adjusts DC offset at the output, and does not have any influence on the PSU voltage values.
Best,
nAr