Hi!
Last time I got JVC amp with Gm VR input circuit which is in fact ngnfb low noise input preamp.
Whole amp sounds wonderful with Magnepans.
Further I wanted to learn more about it and made LtSpice sims of the input and power amp.
For Gm VR (and power section too) I have found that it has ~10MHz knee which is impossible for me to tame.
I tried to put the caps from bases to collectors of all of the transistors, in vain; the knee or even the pitch still existed.
I have changed the types of the transistors from original Japanese to BC550/560 but the oscillations were unchanged so I guess this is up to circuitry, not the transistors type.
THD of the pre is excellent and at 1kHz 40Vpp at the output spice shows ~0.001% and HF harmonics are nicely descending. More, there is soft clipping with this pre.
R18,19 in real is a pot with grounded CT.
Just for the knowledge, please advise what techniques should I use to fight those ~10MHz peaks (even if they are nonexistent in real pre)?
Thank you in advance for any suggestions!
Last time I got JVC amp with Gm VR input circuit which is in fact ngnfb low noise input preamp.
Whole amp sounds wonderful with Magnepans.
Further I wanted to learn more about it and made LtSpice sims of the input and power amp.
For Gm VR (and power section too) I have found that it has ~10MHz knee which is impossible for me to tame.
I tried to put the caps from bases to collectors of all of the transistors, in vain; the knee or even the pitch still existed.
I have changed the types of the transistors from original Japanese to BC550/560 but the oscillations were unchanged so I guess this is up to circuitry, not the transistors type.
THD of the pre is excellent and at 1kHz 40Vpp at the output spice shows ~0.001% and HF harmonics are nicely descending. More, there is soft clipping with this pre.
R18,19 in real is a pot with grounded CT.
Just for the knowledge, please advise what techniques should I use to fight those ~10MHz peaks (even if they are nonexistent in real pre)?
Thank you in advance for any suggestions!
Attachments
You'll find the 10MHz+ peaks disappear if you take R5, R13 to the supply rails rather than to the emitters of Q3,Q10. The enhanced ( aka Hawksford) cascodes are the culprits.
<edit> You will have to fiddle with the DC levels somewhat as those 1k emitter resistors have about 8V across them. I tested this with 8V voltage sources in series with the 2k7s.
<edit> You will have to fiddle with the DC levels somewhat as those 1k emitter resistors have about 8V across them. I tested this with 8V voltage sources in series with the 2k7s.
Last edited:
It is 60MHz, and it is only "cosmetical" issue, overal gain in peak is about -20dB, so no chance for real instability..
Hi.
Add parallel to the base-emitter of transistors (Q3,Q10) generators current capacitors, the size of 10 - 100 nF.
Used .step param c list 1pF 10nF 100nF
Add parallel to the base-emitter of transistors (Q3,Q10) generators current capacitors, the size of 10 - 100 nF.
Used .step param c list 1pF 10nF 100nF
a few places on this board have hinted at open loop transconductance stages with a "variable" resistive load acting as line stage + volume control.
there is fruit on that tree ...
mlloyd1
there is fruit on that tree ...
mlloyd1
R24 fallen in disgrace ? 😀
This "pre" is a V-I converter. The transconductance is determined by R19+R17 (and a little bit by R1//R9) ,reverse value (resistor= V/A to A/V)
Gain is then transconductance x load (R18//R20//R24).Here R18 is that small you can forgett the others.
Mona
Attachments
Thank you!
of course Mjona was right, this is V to I converter used in this particular case for the input stage and input voltage regulator with minimal noise to main poweramp.
proposed caps are working OK but still peaking with tho vol set at min...
seems that the caps should have the values depending on potentiometer's position?
Hi Mjona!
R18,19 are constituting a potentiometer with grounded CT.
The gain is variable and this makes hard to find flexible bandwidith limitating caps.
- Status
- Not open for further replies.