All,
This post is a NOS tidbit. The 7591, 7868, and 6GM5 are electrically equivalent. The difference is the base: Octal, Novar, and Noval, respectively. NOS of all 3 types is scarce, but 6GM5s seem to be the most plentiful. So, if the urge to try NOS in this project can't be controlled, debug with current production 7591s. Then acquire a set of 6GM5s and build adapters from 9 pin mini sockets and Octal plugs.
This post is a NOS tidbit. The 7591, 7868, and 6GM5 are electrically equivalent. The difference is the base: Octal, Novar, and Noval, respectively. NOS of all 3 types is scarce, but 6GM5s seem to be the most plentiful. So, if the urge to try NOS in this project can't be controlled, debug with current production 7591s. Then acquire a set of 6GM5s and build adapters from 9 pin mini sockets and Octal plugs.
John,
SY's remark about stray and Miller capacitances interacting with the NFB resistor pair rings true. Changing to a 1 KOhm grid leak resistor on the non-inverting triode and making the feedback resistor 10 KOhms addresses that concern, while keeping the arbitrary 9% of O/P voltage feedback value. Keeping the FB resistor non-inductive can't hurt as insurance against phase shift trouble.
FWIW, this change is going into "El Cheapo" too.
SY's remark about stray and Miller capacitances interacting with the NFB resistor pair rings true. Changing to a 1 KOhm grid leak resistor on the non-inverting triode and making the feedback resistor 10 KOhms addresses that concern, while keeping the arbitrary 9% of O/P voltage feedback value. Keeping the FB resistor non-inductive can't hurt as insurance against phase shift trouble.
FWIW, this change is going into "El Cheapo" too.
Eli,
I'll make the change, along with other corrections and addition, and draw both types of buffer (for those who don't want to use the extra rails).
While I draw the "popular" power supply, how were you intending to derive the -C ?
Do you recommend the CCS from "El cheapo", or will any reasonable one do?
___________
SY,
Yes.
I'll make the change, along with other corrections and addition, and draw both types of buffer (for those who don't want to use the extra rails).
While I draw the "popular" power supply, how were you intending to derive the -C ?
Do you recommend the CCS from "El cheapo", or will any reasonable one do?
___________
SY,
Yes.
Eli,
I'll make the change, along with other corrections and addition, and draw both types of buffer (for those who don't want to use the extra rails).
While I draw the "popular" power supply, how were you intending to derive the -C ?
Do you recommend the CCS from "El cheapo", or will any reasonable one do?
SY,
Yes.
I'll make the change, along with other corrections and addition, and draw both types of buffer (for those who don't want to use the extra rails).
While I draw the "popular" power supply, how were you intending to derive the -C ?
Do you recommend the CCS from "El cheapo", or will any reasonable one do?
SY,
Yes.
John,
You're setting a regulated B+ rail up. Jim McShane taught me that regulated B+ MUST be paired with a regulated grid bias supply. I hope what follows meets with your approval. Start with a 500 mA. 15 VAC winding, perhaps a "wall wart". Rectify with a "full wave" voltage doubler made from a pair of Schottky diodes. 150 muF. 'lytics in the doubler stack should be adequate. The "raw" DC feeds a LM317 adjustable regulator. IIRC, Avery Fisher used 25 VDC supplies in his 7591 receivers to power phono stage heaters and provide the grid bias of the "finals". The doubler supply I outlined should be quite capable of cutting the 7591s off and supplying the current needed when the 7591 control grids are near zero V.
All,
The simplistic FET buffer that generates a LOT of heat in its load resistor pairs WELL with the choke I/P filter PSU I outlined previously. The FETs turn on instantaneously and the buffers draw more than Icrit. A bleeder resistor is not needed.
You're setting a regulated B+ rail up. Jim McShane taught me that regulated B+ MUST be paired with a regulated grid bias supply. I hope what follows meets with your approval. Start with a 500 mA. 15 VAC winding, perhaps a "wall wart". Rectify with a "full wave" voltage doubler made from a pair of Schottky diodes. 150 muF. 'lytics in the doubler stack should be adequate. The "raw" DC feeds a LM317 adjustable regulator. IIRC, Avery Fisher used 25 VDC supplies in his 7591 receivers to power phono stage heaters and provide the grid bias of the "finals". The doubler supply I outlined should be quite capable of cutting the 7591s off and supplying the current needed when the 7591 control grids are near zero V.
All,
The simplistic FET buffer that generates a LOT of heat in its load resistor pairs WELL with the choke I/P filter PSU I outlined previously. The FETs turn on instantaneously and the buffers draw more than Icrit. A bleeder resistor is not needed.
John,
"El Cheapo" shamelessly copies Bandersnatch's "Merlin" CCS topology, but uses TO92 case FETs instead of TO220 case FETs.
Just about any CCS design that puts a high AC impedance under the splitter/driver's tail should be fine. FETs, BJTs, and tubes can do the job. Anyone whose "sand" tolerance has been exhausted by the MOSFET buffers could use a 6AU6 in the CCS role.
"El Cheapo" shamelessly copies Bandersnatch's "Merlin" CCS topology, but uses TO92 case FETs instead of TO220 case FETs.
Just about any CCS design that puts a high AC impedance under the splitter/driver's tail should be fine. FETs, BJTs, and tubes can do the job. Anyone whose "sand" tolerance has been exhausted by the MOSFET buffers could use a 6AU6 in the CCS role.
Eli,
The high current FET solution does seem best for swinging choke users. Maybe I'll go there after all..
I lost sight of the ball, thinking that the basic design needed -150v, of course it doesn't. The bias supply is easy, and best left unregulated where +B isn't regulated, so that the quiescent current will remain roughly constant as they "track". Certainly this method leads to rock steady current in my 845 amp.
Attached is the latest iteration of the amp signal cct, I notice that "El Cheapo has a 10K in the LTP tail, do we need a resistor?
The high current FET solution does seem best for swinging choke users. Maybe I'll go there after all..
I lost sight of the ball, thinking that the basic design needed -150v, of course it doesn't. The bias supply is easy, and best left unregulated where +B isn't regulated, so that the quiescent current will remain roughly constant as they "track". Certainly this method leads to rock steady current in my 845 amp.
Attached is the latest iteration of the amp signal cct, I notice that "El Cheapo has a 10K in the LTP tail, do we need a resistor?
John,
The 10 KOhm Carbon resistor in the tail of the "El Cheapo" splitter/driver performs 2 functions. It thermally isolates the CCS from the 'T7 cathodes and it holds dissipation down in the TO92 case FETs (360 mW. max., IIRC). If you build with the TO220 case FETs as Bandersnatch did, dissipation is not an issue. A 100 Ohm Carbon part should be adequate, if thermal isolation is needed. "El Cheapo" uses the TO92 case FETs as a cost cutting measure. The name says it all; "El Cheapo" is a CHEAPSKATES' project.
A nit pick, connect Q2's gate to V6's plate. Also, put stoppers on the gates of Q1 and Q2. You might as well be consistent and use 1 KOhm CC parts again.
The 10 KOhm Carbon resistor in the tail of the "El Cheapo" splitter/driver performs 2 functions. It thermally isolates the CCS from the 'T7 cathodes and it holds dissipation down in the TO92 case FETs (360 mW. max., IIRC). If you build with the TO220 case FETs as Bandersnatch did, dissipation is not an issue. A 100 Ohm Carbon part should be adequate, if thermal isolation is needed. "El Cheapo" uses the TO92 case FETs as a cost cutting measure. The name says it all; "El Cheapo" is a CHEAPSKATES' project.
A nit pick, connect Q2's gate to V6's plate. Also, put stoppers on the gates of Q1 and Q2. You might as well be consistent and use 1 KOhm CC parts again.
John,
220 KOhms for the 7591 grid leak resistors is too big. 100 KOhms or less seems on target.
You seem to be moving in the direction of a simpler PSU, with fewer rails. The voltage doubler I previously described has both the V. and the mA. to serve the needs of both the splitter CCSes and 7591 grid bias. "Axe" the regulator IC and follow the doubler stack with a LC section.
John, beat me to the punch with schematic changes. The negative supply comments are still valid.
220 KOhms for the 7591 grid leak resistors is too big. 100 KOhms or less seems on target.
You seem to be moving in the direction of a simpler PSU, with fewer rails. The voltage doubler I previously described has both the V. and the mA. to serve the needs of both the splitter CCSes and 7591 grid bias. "Axe" the regulator IC and follow the doubler stack with a LC section.
John, beat me to the punch with schematic changes. The negative supply comments are still valid.
John,
There is absolutely no need for a NTC device when damper diodes are used for rectification. 6AU4s take the better part of 1 minute to warm up. Cathode stripping is NEVER an issue.
I think in terms of connecting the check valve decoupling networks to the fully filtered 7591 B+. Also, the decoupling resistor is too large for that hookup. I appreciate that the hookup I'm suggesting requires 150 mA. 2nd inductors. I'm concerned that ripple levels will be too high.
Unfortunately for UK and other "230" V. country residents, the Hammond 715 is "115" V. only. Americans and Canadians are fine.
There is absolutely no need for a NTC device when damper diodes are used for rectification. 6AU4s take the better part of 1 minute to warm up. Cathode stripping is NEVER an issue.
I think in terms of connecting the check valve decoupling networks to the fully filtered 7591 B+. Also, the decoupling resistor is too large for that hookup. I appreciate that the hookup I'm suggesting requires 150 mA. 2nd inductors. I'm concerned that ripple levels will be too high.
Unfortunately for UK and other "230" V. country residents, the Hammond 715 is "115" V. only. Americans and Canadians are fine.
Eli,
I have a vested interest in 100mA chokes as I own 2.
-C Ripple prediction is 30mV
What I'm wondering about is if the 20uF for +B1 holds enough energy...
Silly me. I've got 2 designs in my brain. I'll pull it.
+B2 Ripple prediction is 100mV at quiescent, 200mV at max output current. There is only a small cap in the filter at present.
I have a vested interest in 100mA chokes as I own 2.
-C Ripple prediction is 30mV
What I'm wondering about is if the 20uF for +B1 holds enough energy...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.