Hello,
Looks to me like that the cart is ahead of the horse. That Constant Current Sink or Choke should be ahead of the tube. Used as a Constant Current sink all the benefits of increased Power Supply Rejection Ratio and reduced distortion due to holding mu, gm and rp relatively constant will be lost. Spool it up plug it in and run a FFT both as a Constant Current Source and Constant Current Sink. Which one has lower noise and lower distortion? While you are at it, which one will swing more current into the load?
DT
All Just for fun!
Looks to me like that the cart is ahead of the horse. That Constant Current Sink or Choke should be ahead of the tube. Used as a Constant Current sink all the benefits of increased Power Supply Rejection Ratio and reduced distortion due to holding mu, gm and rp relatively constant will be lost. Spool it up plug it in and run a FFT both as a Constant Current Source and Constant Current Sink. Which one has lower noise and lower distortion? While you are at it, which one will swing more current into the load?
DT
All Just for fun!
I think you can make this a lot simpler and improve the performance.
A few questions:
1. Are you committed to not having a feedback loop? With a pentode output stage, the source impedance will be high as will the LF distortion.
2. Is there a reason you're not direct coupling the voltage amp and the phase splitter?
3. Do you want to be able to adjust the idle current in the output tubes?
4. Are you allergic to grid stopper resistors?
1. The feedback loop is absolutely necessary with the 807s, I just haven't quite figured out the best way to apply it. Can I use the grounded grid of the LTP? I know how to measure the open loop gain and the closed loop gain to compute the amount of feedback but I am not sure where to start with the values for the feedback resistors.
2. Only for simplicity at this stage of the build. This is really a "breadboard" and learning experience for me. I wanted to try the grid choke/CF driver stage and the Bogen amp had the choke. Unfortunately, the power transformer was bad. The replacement I found in my junk box from an Altec has insufficient B+ for the 807s. I may change the OP tubes to 300Bs with self bias. That would be the easy way out of course, as it would mitigate the need for GNFB also.
3. Yes, I do need to adjust the current in the OP tubes. At the moment I either need to reduce the grid bias or increase the B+ to make the 807s happy.
4. No, I have been lazy about installing them. Again, this is a breadboard or prototype and I am trying to get the circuit topology to work with the parts I have on hand.
Thanks for your input!
2. Only for simplicity at this stage of the build. This is really a "breadboard" and learning experience for me. I wanted to try the grid choke/CF driver stage and the Bogen amp had the choke. Unfortunately, the power transformer was bad. The replacement I found in my junk box from an Altec has insufficient B+ for the 807s. I may change the OP tubes to 300Bs with self bias. That would be the easy way out of course, as it would mitigate the need for GNFB also.
3. Yes, I do need to adjust the current in the OP tubes. At the moment I either need to reduce the grid bias or increase the B+ to make the 807s happy.
4. No, I have been lazy about installing them. Again, this is a breadboard or prototype and I am trying to get the circuit topology to work with the parts I have on hand.
Thanks for your input!
No gridstoppers = lots of grief. Especially on a breadboard.
I'd probably move the input stage closer to a Mullard 5-20, i.e., direct couple the input tube to the long tail pair. That gets rid of several components and stabilizes things. The best point to return your feedback is to the cathode of the input tube.
As an alternative, you could also insert another voltage amp before the cathode follower, then return feedback from the output tube plates to the driver plates or cathodes.
I'd probably move the input stage closer to a Mullard 5-20, i.e., direct couple the input tube to the long tail pair. That gets rid of several components and stabilizes things. The best point to return your feedback is to the cathode of the input tube.
As an alternative, you could also insert another voltage amp before the cathode follower, then return feedback from the output tube plates to the driver plates or cathodes.
I did a project that used 807s. Vixen Main Schemo. 807s tend to produce more higher order harmonics, and will need the added assistance of local NFB. This was arranged per O. Schade's suggestion: feed back 10% of the AC plate voltage to the input node. In this case, the input to the grid drivers. This definitely helps mitigate the open loop pentode nastiness and tighten up the bass. It still doesn't quite get there, but about 6db(v) of gNFB takes care of cleaning up the sonics. You don't want much more than that, otherwise, the sound becomes "solid statey", and that's not good.
This is an older design, and I'd've done it a bit different today, but it sounds so good that I see no reason to change it now.
Here's how to implement gNFB with an LTP: Le Renard Main Schemo. Given that the LTP control grid is at DC ground, it's especially easy to make the gNFB variable. This method is just the same as that used for SS amps.
Seems to me... (just thinking out loud here)
[1] that a cathode-follower, utilizing the unavoidable series resistance of an inductor to achieve the grid bias operating point ... is a good idea in general, but needs to specify inductors carefully (i.e. not any old 22 to 100 H inductor will do).
[2] And... that if one is going to invest in the "right" inductor (whatever that turns out to be), then why not invest in a similar, "right" transformer, in step-up mode? Put it in the same position as the inductor, and get some voltage amplification along with the usual power amplification. I mean, it probably isn't the case that the next stage needs a 1.2K to 4K input impedance.
[3] Isn't it also the case that there are good, "cheap" current sources that one can make rather trivially these days with MOSFETS? I mean... we're only talking about a few milliamps here ... 8±5 ma? ... and for setpoint voltage between 0-40V, power dissipation doesn't even appreciably exceed 300 milliwatts ... the usual dissipation limit for a TO92 can.
Maybe I'm missing something important. Like the elegance and striking beauty of using a hunk of iron and bunch of copper to effect an A/C signal path hi-Z doohickey. I just like using modern silicon for some things, is all. Ah well, compromised in the noggin.
GoatGuy
[1] that a cathode-follower, utilizing the unavoidable series resistance of an inductor to achieve the grid bias operating point ... is a good idea in general, but needs to specify inductors carefully (i.e. not any old 22 to 100 H inductor will do).
[2] And... that if one is going to invest in the "right" inductor (whatever that turns out to be), then why not invest in a similar, "right" transformer, in step-up mode? Put it in the same position as the inductor, and get some voltage amplification along with the usual power amplification. I mean, it probably isn't the case that the next stage needs a 1.2K to 4K input impedance.
[3] Isn't it also the case that there are good, "cheap" current sources that one can make rather trivially these days with MOSFETS? I mean... we're only talking about a few milliamps here ... 8±5 ma? ... and for setpoint voltage between 0-40V, power dissipation doesn't even appreciably exceed 300 milliwatts ... the usual dissipation limit for a TO92 can.
Maybe I'm missing something important. Like the elegance and striking beauty of using a hunk of iron and bunch of copper to effect an A/C signal path hi-Z doohickey. I just like using modern silicon for some things, is all. Ah well, compromised in the noggin.
GoatGuy
In fact there are cheap, nice depletion mode MOSFETs that serve the purpose well: DF2535, which is almost perfect in specs. Tie gate to ground, put source-resistor in there at R = 2500 / milliamps (desired) ... and you're good to go. Very solid. A P-type enhancement mode could also be used, reversing source and drain. Mouser lists for $0.70 apiece.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.