Well, this is interesting.
I placed a .0068uF 3kV ceramic capacitor to ground on each side of the ac feed to the doubler, because I was concerned there may have been a lot of HF hash going back into the transformer.
I also replaced the 22k dropper that feeds the input tube anode supply with 8.2k.
The result is quite surprising in the HF end. Much better detail and imaging, even though I am listening in a small room. Now as to which mod (or both) that produced this, I don't know.
I now have 159v and 74v on the plates of the 6N1P, so maybe a little more headroom but I haven't altered the current much. That is my next goal, and the reason for changing the dropper value.
Yay! I just found out they are opening a Jaycar store in my town. That means a better supply of parts without ordering from away.
Gary
I placed a .0068uF 3kV ceramic capacitor to ground on each side of the ac feed to the doubler, because I was concerned there may have been a lot of HF hash going back into the transformer.
I also replaced the 22k dropper that feeds the input tube anode supply with 8.2k.
The result is quite surprising in the HF end. Much better detail and imaging, even though I am listening in a small room. Now as to which mod (or both) that produced this, I don't know.
I now have 159v and 74v on the plates of the 6N1P, so maybe a little more headroom but I haven't altered the current much. That is my next goal, and the reason for changing the dropper value.
Yay! I just found out they are opening a Jaycar store in my town. That means a better supply of parts without ordering from away.
Gary
I have quickly scanned the thread and enjoyed the enthusiasm evident in what you guys are doing!
Slightly off-topic (not sure where else to ask the question) ......
Hope SY is reading here!
I also read the references to SY's articles, but could not find reference there to: Why a number of LEDs instead of a suitable zener diode? (or in this case 2 x 5,6V per stage for those worried about the temp-co)?
Then just a remark re screen resistors in UL, which came up a few pages ago:
One must remember that, different from control grids, they draw current. Thus a screen grid will have a finite input impedance. Inserting too large a resistor between G2 and its UL tap will influence the characteristics of the stage. Being fed from a fairly low impedance (the tap on the transformer), the impedance G2 sees will play a role. (This apart from the resistor performing duty as a stopper.) As far as I know only EL34s work better with significant series resistors for G2 (usually 1K). Rare distortion graphs depicting both modes show an audible difference.
Not sure what the effect is with other power tubes - also off-topic here; but something worth considering generally.
Back to topic.
Slightly off-topic (not sure where else to ask the question) ......
Hope SY is reading here!
I also read the references to SY's articles, but could not find reference there to: Why a number of LEDs instead of a suitable zener diode? (or in this case 2 x 5,6V per stage for those worried about the temp-co)?
Then just a remark re screen resistors in UL, which came up a few pages ago:
One must remember that, different from control grids, they draw current. Thus a screen grid will have a finite input impedance. Inserting too large a resistor between G2 and its UL tap will influence the characteristics of the stage. Being fed from a fairly low impedance (the tap on the transformer), the impedance G2 sees will play a role. (This apart from the resistor performing duty as a stopper.) As far as I know only EL34s work better with significant series resistors for G2 (usually 1K). Rare distortion graphs depicting both modes show an audible difference.
Not sure what the effect is with other power tubes - also off-topic here; but something worth considering generally.
Back to topic.
Err no, there shouldn't have been any mention of UL taps, as this amp has no UL connections. We were discussing screen resistors though, and they are taken straight to B+.
Well, it appears that it is going to be difficult (read: impossible for me) to get enough current through the first two stages given the B+ available. I took the first stage up to about 3mA and it looked like I was going to start to run out of voltage to swing on the plate, and that was before starting to ramp up the splitter current. I don't think I could notice any improvement over my previous iteration, but I hadn't altered the splitter so there could have been more to be had there.
So it looks like Ian's tripler idea is the best bet so far. I would still like to see some more current than Ian is using, as all the reports I have read recommend 4mA or more through the 6N1P to make it sound the best. At least with the 350v supply, we can keep the load resistors from getting too low. I was down to 47k with a cathode resistor of 270 ohms on the first stage. I don't know enough to know how low is too low.
If I do go the tripler route, the PCB will be thrown out as I am sick of pulling the tubes and PCB every time I want to make some changes.😡
Gary
So it looks like Ian's tripler idea is the best bet so far. I would still like to see some more current than Ian is using, as all the reports I have read recommend 4mA or more through the 6N1P to make it sound the best. At least with the 350v supply, we can keep the load resistors from getting too low. I was down to 47k with a cathode resistor of 270 ohms on the first stage. I don't know enough to know how low is too low.
If I do go the tripler route, the PCB will be thrown out as I am sick of pulling the tubes and PCB every time I want to make some changes.😡
Gary
Hi Johan,
Thanks for joining in. You are welcome. I don't think your post was off-topic, the whole thread is off-topic, its supposed to be about the schematic but we nailed that and went onto modifications. The zeners are an interesting idea, I have only ever heard of them being used in conjunction with a CCS and bypass cap to help overload recovery. Need to think about it.
With SY's LED's, the RLD uses one bank per channel, I am using 4 seperate banks at the moment and they can be easily paralelled to be the same as the RLD, I have not tried that yet. I tested the whole 72 LED's and arranged them in matched strings, the bias on each output tube is 33.2mA/10.62V, 33.3mA/10.55V on one channel and 34.6mA/10.56V, 34.4mA/10.63V on the other, when I first checked it. With the RLD the regulated screen voltage would be adjusted to adjust the bias, I am just using fixed screen resistors but would like to build the screen regulators at some stage.
tvr - the grab bag of 100 LED's may be fine, what you want is the old cheap low-brightness LED's, that stay around 1.8V or so for currents of say 8 to 15mA. You want a dynamic impedance around 10 ohms or less, I think this can be approximated by measuring the voltage of the LED at say 8mA and 15mA, then the dynamic impedance = (change in voltage)/(change in current). For example if you get 1.78V @ 8mA, and 1.84V @ 15mA, dynamic impedance = (1.84-1.78)/(15-8) = 8.57 ohms, no problems with that LED. This is a simple way of looking at it and may not be necessarily correct, the correct way to do it is explained in the RLD article. As you said, any LED would do for experimental purposes, mate, with mains caps for coupling caps, cheap resistors, four dollar output tubes, luck of the day output trannies, you name it we got it, and the poor little amp still surprises me with clarity and guts 🙂
Gary, the cathode resistor on the input stage needs to give enough bias to prevent the input signal from your source from clipping, so you might want at least 0.8V DC or more. You won't get much more voltage across the valve by lowering the cathode resistor, you really only have the anode resistor (in series with the internal resistance of the valve - plate resistance) and the B+ to play with. The whole bias thing throughout the amp seems to be a much more delicate/complex issue than I ever thought before e.g. getting the signal to clip evenly, balancing the gain, distortion and output impedance of each stage. Then keeping the phase within whatever is reasonable and the output currents balanced within each OPT.
We are looking at around 380V max for driver stage with the tripler. On my amp the driver stage appears to be clipping before the output stage, there's a lot of voltage gain (around x 22) I think with the 220K anode load on the first stage, but I need to read up on it. Will play this weekend. Thanks for the tip on the ceramic caps in the power supply. If you mount all the components on the wrong side of the PCB you will find it easy to change components. The tripler is only one cap and one diode addition, no reconstruction required. Whatever your current B+ is, the tripler will add almost half again. We can also add a tripler below the upper tripler, to get a minus 130V relative to the current ground, might be useful if regulated and adjusted for a fixed bias setup, or for a long tailed pair phase splitter with or without CCS. The LED array took up surprisingly little space Gary, that is, with 3mm LED's. One (dual) array for one channel shown in the pic, smaller than a BIC lighter.
Ian.
Thanks for joining in. You are welcome. I don't think your post was off-topic, the whole thread is off-topic, its supposed to be about the schematic but we nailed that and went onto modifications. The zeners are an interesting idea, I have only ever heard of them being used in conjunction with a CCS and bypass cap to help overload recovery. Need to think about it.
With SY's LED's, the RLD uses one bank per channel, I am using 4 seperate banks at the moment and they can be easily paralelled to be the same as the RLD, I have not tried that yet. I tested the whole 72 LED's and arranged them in matched strings, the bias on each output tube is 33.2mA/10.62V, 33.3mA/10.55V on one channel and 34.6mA/10.56V, 34.4mA/10.63V on the other, when I first checked it. With the RLD the regulated screen voltage would be adjusted to adjust the bias, I am just using fixed screen resistors but would like to build the screen regulators at some stage.
tvr - the grab bag of 100 LED's may be fine, what you want is the old cheap low-brightness LED's, that stay around 1.8V or so for currents of say 8 to 15mA. You want a dynamic impedance around 10 ohms or less, I think this can be approximated by measuring the voltage of the LED at say 8mA and 15mA, then the dynamic impedance = (change in voltage)/(change in current). For example if you get 1.78V @ 8mA, and 1.84V @ 15mA, dynamic impedance = (1.84-1.78)/(15-8) = 8.57 ohms, no problems with that LED. This is a simple way of looking at it and may not be necessarily correct, the correct way to do it is explained in the RLD article. As you said, any LED would do for experimental purposes, mate, with mains caps for coupling caps, cheap resistors, four dollar output tubes, luck of the day output trannies, you name it we got it, and the poor little amp still surprises me with clarity and guts 🙂
Gary, the cathode resistor on the input stage needs to give enough bias to prevent the input signal from your source from clipping, so you might want at least 0.8V DC or more. You won't get much more voltage across the valve by lowering the cathode resistor, you really only have the anode resistor (in series with the internal resistance of the valve - plate resistance) and the B+ to play with. The whole bias thing throughout the amp seems to be a much more delicate/complex issue than I ever thought before e.g. getting the signal to clip evenly, balancing the gain, distortion and output impedance of each stage. Then keeping the phase within whatever is reasonable and the output currents balanced within each OPT.
We are looking at around 380V max for driver stage with the tripler. On my amp the driver stage appears to be clipping before the output stage, there's a lot of voltage gain (around x 22) I think with the 220K anode load on the first stage, but I need to read up on it. Will play this weekend. Thanks for the tip on the ceramic caps in the power supply. If you mount all the components on the wrong side of the PCB you will find it easy to change components. The tripler is only one cap and one diode addition, no reconstruction required. Whatever your current B+ is, the tripler will add almost half again. We can also add a tripler below the upper tripler, to get a minus 130V relative to the current ground, might be useful if regulated and adjusted for a fixed bias setup, or for a long tailed pair phase splitter with or without CCS. The LED array took up surprisingly little space Gary, that is, with 3mm LED's. One (dual) array for one channel shown in the pic, smaller than a BIC lighter.
Ian.
Found, 100 3mm red LED's on auction. Actually I got 200 for a delivered price under $10. ( I am making the reach here that it will be great and I will want to fit them into my HK A300)
I do declare you're right. Must have slipped in through the cracks (of my mind, that is). Advice then to be put in the deep-freeze.
Ian444,
No fine. Zeners are supposed to be noisy (well, they actually are), but I have not found that to be a factor as bias source in power amplifier bias. One bypasses with a C anyway. Also not adjustable, but an easy cheap source of fixed bias.
No fine. Zeners are supposed to be noisy (well, they actually are), but I have not found that to be a factor as bias source in power amplifier bias. One bypasses with a C anyway. Also not adjustable, but an easy cheap source of fixed bias.
With LED's on the way from E-Bay, I ordered new tubes from Tube-Man and all the passive parts from Mouser. Total cost is in the $55 range.
Sounds good tvr. I tried a few more things the last couple of days, except Gary's power supply cap mod, I need to buy some caps.
The first stage of the input stage has plenty of clean gain, but the cathodyne splitter output is limited to around 8V rms before clipping. The peaks of the sine wave get flatter and the troughs get narrower on the anode of the splitter, the cathode output of the splitter does not suffer so much. Thought it may be caused by grid current flowing in the output tubes so I tried bigger stoppers but this did not help and reduced the sound quality.
I then tried different resistance values on the first stage anode and on the cathodyne. No noticeable benefits were realized, so I just left the new values in there.
I tied the cathodes together and ran two LED arrays in parallel. This did not give any noticeable sound quality gains and made biasing the output tubes more difficult, so I put it back to one LED array per cathode. The LED arrays seem to have shifted in their characteristics, the currents have changed for similar cathode voltages, I swapped the tubes around and got the DC currents within 1 or 2mA for each channel, but one channel is running at 44mA and the other at 38mA so I need to build the screen regulator for each channel, or have a look to see if I have more tubes that may match better. This is a big change from when I first fitted the LED arrays.
The power output is still low at around 6W into 8 ohms and 3.5W into 5 ohms. Voltage swing on the OPT is 85V rms, with 7.6V rms feeding the output tube grids (with 8 ohm load). I'm a bit confused with this, why less power into 5 ohms? Because the impedance of the OPT primary is halved allowing smaller voltage swing? 85V on the OPT seems low? Is the primary impedance too low even for 8 ohms and this is limiting the output power? I'm missing some basic understanding here, big time... We have 3 amp owners on this thread all reporting approx 6W output, is it the OPT's, is it the low gm of 6P1P compared to EL84?
It looks like I can remove the heater biasing cct if I read the 6N1P-EV datasheet correctly. It says Ukh is +100, -250V. So it should be no problem, if I am reading it correctly. How do you interpret this, cathode can be 100V or 250V above the heater potential? I take it as 250V but it seems ambiguous. The Svetlana 6N1P is only rated +/- 100V. I originally fitted the bias to run some 12AT7, will stick with 6N1P.
So that's about it from me, close to calling it a day. Just need to find some answers to questions in due course. Current schematic with AC and DC voltages here.
The first stage of the input stage has plenty of clean gain, but the cathodyne splitter output is limited to around 8V rms before clipping. The peaks of the sine wave get flatter and the troughs get narrower on the anode of the splitter, the cathode output of the splitter does not suffer so much. Thought it may be caused by grid current flowing in the output tubes so I tried bigger stoppers but this did not help and reduced the sound quality.
I then tried different resistance values on the first stage anode and on the cathodyne. No noticeable benefits were realized, so I just left the new values in there.
I tied the cathodes together and ran two LED arrays in parallel. This did not give any noticeable sound quality gains and made biasing the output tubes more difficult, so I put it back to one LED array per cathode. The LED arrays seem to have shifted in their characteristics, the currents have changed for similar cathode voltages, I swapped the tubes around and got the DC currents within 1 or 2mA for each channel, but one channel is running at 44mA and the other at 38mA so I need to build the screen regulator for each channel, or have a look to see if I have more tubes that may match better. This is a big change from when I first fitted the LED arrays.
The power output is still low at around 6W into 8 ohms and 3.5W into 5 ohms. Voltage swing on the OPT is 85V rms, with 7.6V rms feeding the output tube grids (with 8 ohm load). I'm a bit confused with this, why less power into 5 ohms? Because the impedance of the OPT primary is halved allowing smaller voltage swing? 85V on the OPT seems low? Is the primary impedance too low even for 8 ohms and this is limiting the output power? I'm missing some basic understanding here, big time... We have 3 amp owners on this thread all reporting approx 6W output, is it the OPT's, is it the low gm of 6P1P compared to EL84?
It looks like I can remove the heater biasing cct if I read the 6N1P-EV datasheet correctly. It says Ukh is +100, -250V. So it should be no problem, if I am reading it correctly. How do you interpret this, cathode can be 100V or 250V above the heater potential? I take it as 250V but it seems ambiguous. The Svetlana 6N1P is only rated +/- 100V. I originally fitted the bias to run some 12AT7, will stick with 6N1P.
So that's about it from me, close to calling it a day. Just need to find some answers to questions in due course. Current schematic with AC and DC voltages here.
Ah, prettier than my white-out and red pen copy. Waiting for parts gives me time to figure out my new e-mu 1616. It's noise is low enough to make amp measurements valid. I had so much 60 and harmonics in my Audigy and mic preamp it was hard to see if I was changing the amp. I want to see if raising the transformers on brass hardware reduces hum any, or if all I bright ideas on power and ground paths is measurable.
I bought a few extra parts so I can fiddle a little. I did not go extreem on the parts, mostly carbon film. Got parts to do separate B2 for R&L, a handful of .01 disks, and some choices to bring a tad of feedback in to the input stage. Fast diodes were so cheap I bought some. I am not sure I believe they make a difference if a decent bypass cap is used, but only by trying it I will know. It should show up on the spectrum analyzer.
Only thing I can guess is the OPT is of the quality we suspect. Random amount of turns and whatever wire was available.
Sounds like one needs to burn in the LED's and then trim each string for the same voltage. Otherwise the lowest drop leg will current hog. I am already thinking of an adjustable low Z buffer or something like that.
I have not watched the PS under full load. Wonder if it collapses? If so, a bigger cap bank with a slow start could give it some dynamic headroom. We have droppers anyway, so some thermisters should not hurt things.
I bought a few extra parts so I can fiddle a little. I did not go extreem on the parts, mostly carbon film. Got parts to do separate B2 for R&L, a handful of .01 disks, and some choices to bring a tad of feedback in to the input stage. Fast diodes were so cheap I bought some. I am not sure I believe they make a difference if a decent bypass cap is used, but only by trying it I will know. It should show up on the spectrum analyzer.
Only thing I can guess is the OPT is of the quality we suspect. Random amount of turns and whatever wire was available.
Sounds like one needs to burn in the LED's and then trim each string for the same voltage. Otherwise the lowest drop leg will current hog. I am already thinking of an adjustable low Z buffer or something like that.
I have not watched the PS under full load. Wonder if it collapses? If so, a bigger cap bank with a slow start could give it some dynamic headroom. We have droppers anyway, so some thermisters should not hurt things.
Spec sheets I have seen for the EV model state the the max heater to cathode is 120v (also applies to the Vi model), the stock 6n1p being 100v.
And typical operation spec is for 7.5 +- 1.5mA plate current.
Here's the one to watch, the EV is derated on the plate voltage to a max of 250v whereas all the other 6n1s are rated to 300v.
The PSU cap mod I did was made tidier now that the cathode resistors are gone. I used one of the vacant pads and tied it to earth. I only picked the cap value because Dick Smith were selling out all their components cheap (there is a Jaycar stockist opening next door to them), and that was one of the few 3kV ones left. My SE amp has 0.01uF 600 polystyrene to ground (factory fitted to the radio that the chassis came from) from the bridge ac to ground.
I'll try and do a power output measurement (haven't even checked that yet) but I tried the amp last night with 5.6 ohm resistors in parallel with my 8 ohm speakers, and the amp didn't like it. Much more distortion.
Gary
Ian, a couple of things I noticed in your circuit:
You have the LEDs lifted above ground by the 4R7 resistors. Didn't SY have his resistors on the tube cathode end? I only ask as although I know it doesn't affect the DC conditions, it may have a bearing on the 0.55v ac ripple you show on the cathode and hence the sonics. I am only going on memory of the RLD circuit though.
I would be interested in your ac signal at the grid of the splitter when you have the other conditions shown in your schematic.
I am seriously tempted to try ECC81s in the front end as I have a couple NOS here. Might be an interesting counterpoint, given the low plate current they are happy with.
I make your plate current to be 1.4mA and 2mA for the 6n1 halves, if I am doing it right with ohms law for the Va/Ra. I still think that is where the most improvement can be made with the 6N1P
I might try that 5R6 in series with the speakers tonight. That might tell us whether the primary impedance of the OPT is off for the 6P1s. Wouldn't surprise me given the sloppy looking windings on my OPTs.
Oh, and check out the little LR8N hv regulator chip if you want to play with screen voltages. Think of it as a 450v LM317.
You have the LEDs lifted above ground by the 4R7 resistors. Didn't SY have his resistors on the tube cathode end? I only ask as although I know it doesn't affect the DC conditions, it may have a bearing on the 0.55v ac ripple you show on the cathode and hence the sonics. I am only going on memory of the RLD circuit though.
I would be interested in your ac signal at the grid of the splitter when you have the other conditions shown in your schematic.
I am seriously tempted to try ECC81s in the front end as I have a couple NOS here. Might be an interesting counterpoint, given the low plate current they are happy with.
I make your plate current to be 1.4mA and 2mA for the 6n1 halves, if I am doing it right with ohms law for the Va/Ra. I still think that is where the most improvement can be made with the 6N1P
I might try that 5R6 in series with the speakers tonight. That might tell us whether the primary impedance of the OPT is off for the 6P1s. Wouldn't surprise me given the sloppy looking windings on my OPTs.
Oh, and check out the little LR8N hv regulator chip if you want to play with screen voltages. Think of it as a 450v LM317.
Oh, and Ian, I meant to ask what you are using to draw your diagrams with? I would like a simple package that has tubes already drawn (mac preferably, but I have WinXP on it as well) to draw circuits up.
Gary, I was using this datasheet here. How did you find the amp after fitting the LM317's as cathode loads?
Re the placement of resistors in a series cct, I don't think it will have any impact on the sonics.
At one stage, with 410mV input, I got 8.64V rms on pin1 and 8.16V rms on pin 6 of the 6N1P. I'm keeping a folder/diary for this amp now as it's too easy for me to forget what I have done.
Thanks for the tip on the LR8N reg, will look into it. I agree about the low plate current in the 6N1P, how do we get 7mA flowing, gonna take a very high B2?
I'm drawing with MSpaint, comes free with every Windoze install. I've had Protel and others, but just keep going back to it.
Re the placement of resistors in a series cct, I don't think it will have any impact on the sonics.
At one stage, with 410mV input, I got 8.64V rms on pin1 and 8.16V rms on pin 6 of the 6N1P. I'm keeping a folder/diary for this amp now as it's too easy for me to forget what I have done.
Thanks for the tip on the LR8N reg, will look into it. I agree about the low plate current in the 6N1P, how do we get 7mA flowing, gonna take a very high B2?
I'm drawing with MSpaint, comes free with every Windoze install. I've had Protel and others, but just keep going back to it.
Here is the datasheet I saw: 6N1P, 6N1PVI, 6N1PEV (6H1n, 6H1nBN, 6H1nEB)
The only way I can see to increase the current is to drop the plate load of the first stage and rebias. I did that as I said, but I didn't want to drop further than 47k until I learned more about the effects in that circuit. I reckon it will take something around 30k of plate load to achieve 4-5mA.
What I don't get with the circuit is this:
If I play around with the first half, I am severely changing plate voltage. What does that do to the splitter, DC wise? If I have 80v on the plate of V1a, that is going to be reflected on grid of V1b. That's fine if the grid/cathode of V1b was going to be in that area anyway (if it was a separate circuit it would settle at that for example). What happens if I leave V1b alone, and rebias V1a so that I get 130v on the plate. Does V1b happily follow? I have a mental block when it comes to DC interactions between stages.
I didn't notice much change sonically with the 317s, but it gives me a stable bias point that I can work from. I still think the output stage is masking any major improvements I make in the driver.
The LR8N looks like a cheap and simple screen regulator. The small package version is limited to 20mA though. At least it's cheap, about 0.80US.
I tried 5.6 in series with the speakers, no difference to running just 8 ohm speakers 😕
I think we need to know the actual load the OPT presents, and that obviously varies from chassis to chassis 🙁
Except for the intermodulation distortion I get on high bass content on my KEF B139 TLs, the amp actually sounds fairly good now. The TLs need an amp with good damping and the Meng doesn't offer that. I certainly find the treble end very easy to listen to in comparison to the Pioneer home theater amp. On acoustic and light electric jazz, the Meng is a clear winner.
I just picked up a Sanyo JCX-2400K receiver for $2 (it actually goes). With its huge heavy chassis , vented wooden case, and modular construction, it lends itself to being tube-converted 😀
Or it would be easy to drop in a T-amp or LM3886 board in place of the STK084 power modules. The only downside is the wire-wrap methods they used.
Looks like another project in the wings.
Gary
OK Ian, I read an interesting post here by Wavebourn in the stereo SE KT88 thread:
"I would suggest anode load of 6N1P 24K, and 360 Ohm in cathode of 6N1P, especially when you drive KT-88 in triode mode. You may use a 3.5V drop bright LED in cathode instead of 360 Ohm resistor."
This circuit uses a B+ of 400v, so you could model fairly closely on that. Note the 3.5v bias, and the 6n1P is on the input.
From what I can tell from the curves, it looks like you will be seeing well over 200v on the plate, which will mean some more jacking of the heaters I think.
Gary
Been thinking about it - 85V rms on the primary of the OPT is nearly a 240V swing peak to peak, and with 10V on the cathode, and B+ of 250V, that's it, hit a brick wall. So nothing left. To get more power, more voltage is needed e.g. EL84 and higher B+. I don't need any more power, but was expecting more, don't know why, maybe because it was advertised as 10W. Maybe biased too hot into class A too. Not to worry. So 6W or less it is for now. Sounds great, no prob.
Gary, seems you and I have both been thinking about the 6N1P about how to get more current flow. I drew a loadline onto the Svetlana datasheet for a 33K anode resistor earlier. Bias 2V, 350V B+. The theory is from valvewizard (still waiting on MJ book). Looks like the anode will sit around 164V under quiescent conditions with 5.7 mA flowing. With a 2V p-p signal the anode will vary between 132 and 190V. 2nd harmonic distortion will be around 5%. Maybe some of that gets cancelled in the PP O/P stage.
Amplification factor u = change in anode V / change in grid V = 58/2 = 29
Anode resistance = change in anode V / change in anode current = 58/0.0017 = 34K
Cathode resistor for 2V bias = 2V/5.7mA = 350 ohms
Voltage gain = u * anode resistor / anode resistor + anode resistance + cathode resistor*(u + 1)
= 29 * 33K / 33K + 34K + 350(29 + 1)
= 12.3
Should be plenty of gain still, I think.
Been thinking about the biasing of the DC-coupled cathodyne. I am guessing the best voltage to bias it at would be when the voltage drops across the anode and cathode resistors are equal under quiescent conditions, to give max linear + and - swing? There are two things happening with an AC signal on the grid 1) the cathode will try to track the grid 2) the tube current varies to give the voltage swings on the outputs. There must be some sort of balance between the two. Still trying to wrap my head around it. Once you build the cct, check the bias under quiescent conditions and draw a loadline and see where the bias sits.
Here is a loadline for the 6N1P with 24K anode resistor, 3.5V bias, 400V B+. The loadline should start at 16.6mA on the LHS, not 16mA, but close enough to get the idea. The 360 ohm bias resistor does not match up with 3.5V on these curves, that would make ~10mA and ~2.2W, right on the max dissipation limit. Sometimes the valves I have do not match the curves very closely either...
Here's a loadline for 230V B+ and 30K anode resistor.
I think tvr asked about power supply sag, I saw the B+ on my amp drop from 254 to 250V at flat chat clipping, it looks clean. The driver tube tripler supply has 1V ripple, can't remember if that is before or after the 33uF caps.
Gary, seems you and I have both been thinking about the 6N1P about how to get more current flow. I drew a loadline onto the Svetlana datasheet for a 33K anode resistor earlier. Bias 2V, 350V B+. The theory is from valvewizard (still waiting on MJ book). Looks like the anode will sit around 164V under quiescent conditions with 5.7 mA flowing. With a 2V p-p signal the anode will vary between 132 and 190V. 2nd harmonic distortion will be around 5%. Maybe some of that gets cancelled in the PP O/P stage.
Amplification factor u = change in anode V / change in grid V = 58/2 = 29
Anode resistance = change in anode V / change in anode current = 58/0.0017 = 34K
Cathode resistor for 2V bias = 2V/5.7mA = 350 ohms
Voltage gain = u * anode resistor / anode resistor + anode resistance + cathode resistor*(u + 1)
= 29 * 33K / 33K + 34K + 350(29 + 1)
= 12.3
Should be plenty of gain still, I think.
Been thinking about the biasing of the DC-coupled cathodyne. I am guessing the best voltage to bias it at would be when the voltage drops across the anode and cathode resistors are equal under quiescent conditions, to give max linear + and - swing? There are two things happening with an AC signal on the grid 1) the cathode will try to track the grid 2) the tube current varies to give the voltage swings on the outputs. There must be some sort of balance between the two. Still trying to wrap my head around it. Once you build the cct, check the bias under quiescent conditions and draw a loadline and see where the bias sits.
Here is a loadline for the 6N1P with 24K anode resistor, 3.5V bias, 400V B+. The loadline should start at 16.6mA on the LHS, not 16mA, but close enough to get the idea. The 360 ohm bias resistor does not match up with 3.5V on these curves, that would make ~10mA and ~2.2W, right on the max dissipation limit. Sometimes the valves I have do not match the curves very closely either...
Here's a loadline for 230V B+ and 30K anode resistor.
I think tvr asked about power supply sag, I saw the B+ on my amp drop from 254 to 250V at flat chat clipping, it looks clean. The driver tube tripler supply has 1V ripple, can't remember if that is before or after the 33uF caps.
I ran the LED array and potential for current hogging past a few engineers at work, and all agreed, likely to be a stability problem. I will know when I get all my parts and try it. It would be nice if they increased their drop when they got hot, then they would self-balance. I know in my notes somewhere I have a nice writeup for a low Z buffer-output voltage regulator. It would allow trim of the bias specific to each tube and be lower impedance than the LED's..
The other thing that I was thinking about was how to compensate for the variance in the OPT. As they seemed to be wound "approximate". The paraphase allowed adjustment of AC amplitude, so by measuring output distortion, best balance could be made. I would guess that is one contributor to why when set with a voltmeter, it was far from optimum. I wonder if a similar adjustment would not be a good idea on the cathodyne? I am probably over-thinking, but waiting for parts does that.
I need to get back to why a single switch and stepped attenuator cause an increase of .05% distortion on my test bench. Cable loop-back, .003%. With switch, .06% Almost no 60 Hz related noise like I had with the Audigy. E-mu is far cleaner. Plenty of measurement room for working on amps.
The other thing that I was thinking about was how to compensate for the variance in the OPT. As they seemed to be wound "approximate". The paraphase allowed adjustment of AC amplitude, so by measuring output distortion, best balance could be made. I would guess that is one contributor to why when set with a voltmeter, it was far from optimum. I wonder if a similar adjustment would not be a good idea on the cathodyne? I am probably over-thinking, but waiting for parts does that.
I need to get back to why a single switch and stepped attenuator cause an increase of .05% distortion on my test bench. Cable loop-back, .003%. With switch, .06% Almost no 60 Hz related noise like I had with the Audigy. E-mu is far cleaner. Plenty of measurement room for working on amps.
- Status
- Not open for further replies.