to do list
Hi,
I implemented the separate transformers for the filaments yesterday afternoon. The hum at the output is larger, but tolerable for testing, without a humpot implemented. When I had the single transformer, offsetting the bias voltages with the static bias adjust pot caused a large and not nullable increase in hum. With the separate transformers there still seems to be an increase although I do not think it is as great. Will try to find a suitable humpot today. I now have 10 ohm resistors to ground on each filament so that the static current can be measured without resorting to measuring anode voltage.
There is a problem with A2 operation. When positive grid is reached there is a large increase in distortion at the 12B4 stage. I still have not done any extensive distortion measurements, but I think it may be greater than I had seen earlier. There may yet be a problem with the bias supply or maybe the 12B4 just doesn't have the poop. Federico, can you model the input impedance of the grids of the 845 when it reaches grid current? I don't know if the distortion increase is a loading issue from the grid or if it is caused by some other circuit feature. I think if I get the time later today I will disconnect the 845s from the circuit, ground the CT of the 1660 and load the secondaries with a couple of kohms to ground to see how the 12B4s react. Make sense?
On the 6SN7, I went back to the 1k pot and 15H inductor on the cathode. I did not want to change to many circuit elements before I have the whole amp tuned in somewhat more. I do think the 6H6 as bias has some real possibilities though. It biased the tube at 5.9V with 360V B+ and 22k Rp. Seems very close to where a person would want it.
Will look for separate caps to ground for the 1660 secondaries. I have not noticed any instability with the 845 though, with or without a cap across the transformer. Also doesn't seem to be much effect (all of this only at 1kHz signal) to placing a resistance across the secondaries. I do have some sort of small damped oscillation that apparently is appearing at the 6SN7. I have ignored it for the moment since as I said it is small (50mV after 12B4), well damped, and at ~500 kHz. Probably due to some breadboard issue. The 845 seems to be highly stable. Just for the heck of it I reversed the feedback connections to give a little positive feedback to the 845 and it didn't complain with any instability.
Anyone have any suggestions for things to try?
Michael
Hi,
I implemented the separate transformers for the filaments yesterday afternoon. The hum at the output is larger, but tolerable for testing, without a humpot implemented. When I had the single transformer, offsetting the bias voltages with the static bias adjust pot caused a large and not nullable increase in hum. With the separate transformers there still seems to be an increase although I do not think it is as great. Will try to find a suitable humpot today. I now have 10 ohm resistors to ground on each filament so that the static current can be measured without resorting to measuring anode voltage.
There is a problem with A2 operation. When positive grid is reached there is a large increase in distortion at the 12B4 stage. I still have not done any extensive distortion measurements, but I think it may be greater than I had seen earlier. There may yet be a problem with the bias supply or maybe the 12B4 just doesn't have the poop. Federico, can you model the input impedance of the grids of the 845 when it reaches grid current? I don't know if the distortion increase is a loading issue from the grid or if it is caused by some other circuit feature. I think if I get the time later today I will disconnect the 845s from the circuit, ground the CT of the 1660 and load the secondaries with a couple of kohms to ground to see how the 12B4s react. Make sense?
On the 6SN7, I went back to the 1k pot and 15H inductor on the cathode. I did not want to change to many circuit elements before I have the whole amp tuned in somewhat more. I do think the 6H6 as bias has some real possibilities though. It biased the tube at 5.9V with 360V B+ and 22k Rp. Seems very close to where a person would want it.
Will look for separate caps to ground for the 1660 secondaries. I have not noticed any instability with the 845 though, with or without a cap across the transformer. Also doesn't seem to be much effect (all of this only at 1kHz signal) to placing a resistance across the secondaries. I do have some sort of small damped oscillation that apparently is appearing at the 6SN7. I have ignored it for the moment since as I said it is small (50mV after 12B4), well damped, and at ~500 kHz. Probably due to some breadboard issue. The 845 seems to be highly stable. Just for the heck of it I reversed the feedback connections to give a little positive feedback to the 845 and it didn't complain with any instability.
Anyone have any suggestions for things to try?
Michael
hi
try to change the res across the 1660 primary frim 80k
to 30k and see if something will change.
It is already ( automatically) modelled since the tube is modelled
as two current generators: one from anode to cath. and the other from grid to cath.
If the curves are o.k. (grid current and anode current) than
all is o.k.
bye
Federico
I will investigate this phenomenon. I never noticed it in simulation.
try to change the res across the 1660 primary frim 80k
to 30k and see if something will change.
It is already ( automatically) modelled since the tube is modelled
as two current generators: one from anode to cath. and the other from grid to cath.
If the curves are o.k. (grid current and anode current) than
all is o.k.
I advice you to use the double cap anyway.
bye
Federico
Michael,
Hi, I’m just catching up on your progress.
I didn’t have any problems with hum with DC filaments on the 845’s.
I’m interested in your bias scheme as this is a place I think I can improve my design. I’m looking at shunt regulators and lowering the series resistance from the bias circuit to help distortion in A2 operation.
Can you show us the schematic you are currently using, this should help with suggestions.
thanks,
Chuck
Hi, I’m just catching up on your progress.
I didn’t have any problems with hum with DC filaments on the 845’s.
I’m interested in your bias scheme as this is a place I think I can improve my design. I’m looking at shunt regulators and lowering the series resistance from the bias circuit to help distortion in A2 operation.
Can you show us the schematic you are currently using, this should help with suggestions.
thanks,
Chuck
bias supply
Hi Chuck,
I've got it mostly drawn up until the last change or two. I'm kind of busy right now, but I will try to do the update and post it sometime this evening.
I've read a lot of posts claiming that DC heaters make the amp sound less good. Tactful enough? Anyway, I thought I would try AC first because it is easiest if nothing else. In some iterations of the circuit the hum has been low enough that you did literally have to stick your ear on the speaker to hear it and I think the speakers I am using are something like 94dB/W/m (old Pioneer HPM 40, feel free to correct the efficiency info). The bias balancing circuit throws a wrench in it though. Every time the pots are changed the hum reappears. I may not have something right yet. As far as I have been able to get it so far though I would say A2 is probably better than clipping, but not too listenable on a full time basis.
Later,
Michael
Hi Chuck,
I've got it mostly drawn up until the last change or two. I'm kind of busy right now, but I will try to do the update and post it sometime this evening.
I've read a lot of posts claiming that DC heaters make the amp sound less good. Tactful enough? Anyway, I thought I would try AC first because it is easiest if nothing else. In some iterations of the circuit the hum has been low enough that you did literally have to stick your ear on the speaker to hear it and I think the speakers I am using are something like 94dB/W/m (old Pioneer HPM 40, feel free to correct the efficiency info). The bias balancing circuit throws a wrench in it though. Every time the pots are changed the hum reappears. I may not have something right yet. As far as I have been able to get it so far though I would say A2 is probably better than clipping, but not too listenable on a full time basis.
Later,
Michael
Quick question
Hello ,
I've been reading this thread with particular interest as I'm just beginning a similar project , maybe even with triode strapped GK71
Been modelling 845's at low volts and find , according to the simulation , find the best compromise for 650V Va to be -75V/80mA with 6.6K A-A optx giving 20 watts class A1 , which would be similar to how you are running yours . However the curve models on the simulation look more like IDHT rather than DHT so I'm taking the results with a pinch of salt The output of 42.75W measured , I gather this is going into A2 . Do you know the maximum measured power before grid current was drawn ?
ps Can't wait to get the project started , 20 pop-can sized motor runs and a 600VA C-core are on the way already and at last I have a use for my 10H 350mA WWII Admiralty chokes !
cheers
316a
audiobot said:
Hello ,
I've been reading this thread with particular interest as I'm just beginning a similar project , maybe even with triode strapped GK71
Been modelling 845's at low volts and find , according to the simulation , find the best compromise for 650V Va to be -75V/80mA with 6.6K A-A optx giving 20 watts class A1 , which would be similar to how you are running yours . However the curve models on the simulation look more like IDHT rather than DHT so I'm taking the results with a pinch of salt The output of 42.75W measured , I gather this is going into A2 . Do you know the maximum measured power before grid current was drawn ? ps Can't wait to get the project started , 20 pop-can sized motor runs and a 600VA C-core are on the way already and at last I have a use for my 10H 350mA WWII Admiralty chokes !
cheers
316a
A1 power
Hi,
Chuck, I haven't redone the schematic yet. I'm still not sure I have this area set up correctly.
316a, I am using the 1620 in the 11kohm configuration at the moment. Since it is something of a pain to change I have not been jumping back and forth on the output impedances, but I should take a look and listen to the 6k config for sure. The power out at onset of grid current seems to be right around 17-18W at 660 B+, -85V bias and 82.5mA. Distortion skyrockets after that point, much more than Federicos model suggests, which leads me to think I may still have something laid out wrong.
GK71 should be an interesting tube. I may try them after I have optimized this circuit, if I can figure out how to socket them temporarily. Those oil caps are big. I have them strewn all over my breadboard now. Must try to minimize number needed or the amp will require it's own room.
Michael
Hi,
Chuck, I haven't redone the schematic yet. I'm still not sure I have this area set up correctly.
316a, I am using the 1620 in the 11kohm configuration at the moment. Since it is something of a pain to change I have not been jumping back and forth on the output impedances, but I should take a look and listen to the 6k config for sure. The power out at onset of grid current seems to be right around 17-18W at 660 B+, -85V bias and 82.5mA. Distortion skyrockets after that point, much more than Federicos model suggests, which leads me to think I may still have something laid out wrong.
GK71 should be an interesting tube. I may try them after I have optimized this circuit, if I can figure out how to socket them temporarily. Those oil caps are big. I have them strewn all over my breadboard now. Must try to minimize number needed or the amp will require it's own room.
Michael
Re: A1 power
Hello ,
The model may apply to NOS original 845 , I assume you're using Chinese 845 so would expect the curves to differ somewhat . GK71 sockets can be manufactured from teflon or ceramic B7a sockets , otherwise use capacitor clips to clamp the valves into position and directly solder the pins . I've already made some from teflon B7A sockets and the pictures should be here in the archives somewhere . From the tests I have carried out it appears triode strapped GK71 can simply be treated as an 845 with mu of 4.5 , various bias points are approximately 20% different between the two , Ra is around 1.8K > 60mA . Transconductance measures higher with both G2 and G3 tied together which may lead to interesting comparisons to G3 tied to 0V . Hopefully I'll start mine early next week , time allowing . Keep up the good work , interesting thread
cheers
316a
audiobot said:
Hello ,
The model may apply to NOS original 845 , I assume you're using Chinese 845 so would expect the curves to differ somewhat . GK71 sockets can be manufactured from teflon or ceramic B7a sockets , otherwise use capacitor clips to clamp the valves into position and directly solder the pins . I've already made some from teflon B7A sockets and the pictures should be here in the archives somewhere . From the tests I have carried out it appears triode strapped GK71 can simply be treated as an 845 with mu of 4.5 , various bias points are approximately 20% different between the two , Ra is around 1.8K > 60mA . Transconductance measures higher with both G2 and G3 tied together which may lead to interesting comparisons to G3 tied to 0V . Hopefully I'll start mine early next week , time allowing . Keep up the good work , interesting thread
cheers
316a
GK-71
Hi,
Good idea with the capacitor clamp. I think I may have some around that will fit. Long ago when I was thinking of making sockets myself I found some female wire crimp type fittings that fit the pins perfectly so I could avoid soldering the pins too. As a matter of interest, have you ever tried the tube with the G2 tied to G1 instead of the plate? I read something in an old datasheet or an early fifties magazine that seemed to think that was the way to go.
Michael
Hi,
Good idea with the capacitor clamp. I think I may have some around that will fit. Long ago when I was thinking of making sockets myself I found some female wire crimp type fittings that fit the pins perfectly so I could avoid soldering the pins too. As a matter of interest, have you ever tried the tube with the G2 tied to G1 instead of the plate? I read something in an old datasheet or an early fifties magazine that seemed to think that was the way to go.
Michael
Re: GK-71
Hello ,
Bullet receptacles (red and blue colour code) , if given a squeeze , fit the pins if the valve is clamped . Another alternative is to clamp the valve and use 15amp and 30A screw terminal blocks but I don't think you use these things in the USA . Better off knocking up a socket by drilling out and refitting a B7A , the teflons need a retainer to hold the contacts but looking at a Johnson knockoff , it looks like these may work just by drilling and prising . here's one I made earlier ...
No idea about connecting G1 and G2 , guess G3 is connected to the cathode . I could always give this a try on the tester and see what happens
cheers
316a
[
audiobot said:
Hello ,
Bullet receptacles (red and blue colour code) , if given a squeeze , fit the pins if the valve is clamped . Another alternative is to clamp the valve and use 15amp and 30A screw terminal blocks but I don't think you use these things in the USA . Better off knocking up a socket by drilling out and refitting a B7A , the teflons need a retainer to hold the contacts but looking at a Johnson knockoff , it looks like these may work just by drilling and prising . here's one I made earlier ...
An externally hosted image should be here but it was not working when we last tested it.
No idea about connecting G1 and G2 , guess G3 is connected to the cathode . I could always give this a try on the tester and see what happens
cheers
316a
[
A2 ruminations
Woe is me, all is lost, oye vay..... I don't think the driver stage can do A2. I tried it out with a 10k resistor spanning the secondary of the 1660 and the output was way down and the distortion was way up. I tried it both with the 1660 secondary CTs connected to the operating bias suppy and also connected together and grounded. I was lucky to get 45Vrms of fairly clean signal out. If a 60k resistor replaced the 10k, it was no problem to get an ultraclean 100Vrms. So I fear when the grid circuit impedance goes down my driver sits down and cries. Seem valid? Or am I mistaken?
If true, this leaves some executive decision making to do. Is 15W enough? Yes-Keep everything as planned and build. No-Option one different driver/driver coupling setup (6B4G or cathode follower direct or transformer coupled). Option two-forget choke input on the power supply and raise the B+ while dropping the bias keep in A1. Others?
Since option two was easy to try with the variable supplies I am using, I tried it. With the bias at -122V and the plate ~820V it was a straight shot to 32W (16Vrms->8ohm) with 3rd order distortion of 0.55% (90mVrms) and all other orders =<5mV.
Opinions?
Michael
Woe is me, all is lost, oye vay..... I don't think the driver stage can do A2. I tried it out with a 10k resistor spanning the secondary of the 1660 and the output was way down and the distortion was way up. I tried it both with the 1660 secondary CTs connected to the operating bias suppy and also connected together and grounded. I was lucky to get 45Vrms of fairly clean signal out. If a 60k resistor replaced the 10k, it was no problem to get an ultraclean 100Vrms. So I fear when the grid circuit impedance goes down my driver sits down and cries. Seem valid? Or am I mistaken?
If true, this leaves some executive decision making to do. Is 15W enough? Yes-Keep everything as planned and build. No-Option one different driver/driver coupling setup (6B4G or cathode follower direct or transformer coupled). Option two-forget choke input on the power supply and raise the B+ while dropping the bias keep in A1. Others?
Since option two was easy to try with the variable supplies I am using, I tried it. With the bias at -122V and the plate ~820V it was a straight shot to 32W (16Vrms->8ohm) with 3rd order distortion of 0.55% (90mVrms) and all other orders =<5mV.
Opinions?
Michael
5998 et al
Hi,
The 5998 looks interesting for sure. A couple of points though. At the 50mA current max for the LL1660 interstage I fear the Rp is approaching that of the 12B4 and my feeling is that it is the plate resistance that hurts. I am only asking the 12B4s to sink a couple of milliamps and the same setup easily gives me 6 or 7W when fed through an output transformer (see thread re 12B4). Actually, if I remember correctly, I could take out 1.5W through the interstage if the load impedance was high. Still, might be worth a try. Think it would be better than a 2A3/6B4G?
Michael
Hi,
The 5998 looks interesting for sure. A couple of points though. At the 50mA current max for the LL1660 interstage I fear the Rp is approaching that of the 12B4 and my feeling is that it is the plate resistance that hurts. I am only asking the 12B4s to sink a couple of milliamps and the same setup easily gives me 6 or 7W when fed through an output transformer (see thread re 12B4). Actually, if I remember correctly, I could take out 1.5W through the interstage if the load impedance was high. Still, might be worth a try. Think it would be better than a 2A3/6B4G?
Michael
Re: 5998 et al
Hello ,
You may be right there , I had no idea about the limitations of the IT . I don't know about 2A3/6B4G , maybe excessive and maybe or maybe not requiring DC filaments adding to complexity . A good possibilty may be the GEC A2134 triode strapped . A2134 has Pdiss of 10 watts , B7G base , Ra 1k , mu of 9 . Very low distortion , cheap and readily available here in the UK and they often pop up on Ebay for peanuts . One point to not regarding 5998 , these may require careful selection to find samples with matched sections , 7236 may be preferable . Of course there's always the 6CK4 which is an octal 12B4a with 10w dissipation...
cheers
316a
audiobot said:
Hello ,
You may be right there , I had no idea about the limitations of the IT . I don't know about 2A3/6B4G , maybe excessive and maybe or maybe not requiring DC filaments adding to complexity . A good possibilty may be the GEC A2134 triode strapped . A2134 has Pdiss of 10 watts , B7G base , Ra 1k , mu of 9 . Very low distortion , cheap and readily available here in the UK and they often pop up on Ebay for peanuts . One point to not regarding 5998 , these may require careful selection to find samples with matched sections , 7236 may be preferable . Of course there's always the 6CK4 which is an octal 12B4a with 10w dissipation...
cheers
316a
valve suggestions
Hey Nic,
You have a wealth of good valve suggestions. Since I like the 12B4 a lot I especially like the 6CK4. Will have to get a few of those to try in a standalone. The question is, for me because I have no idea, can any valve drive a 2.25:2 interstage into grid current? Or does the stepdown need to be larger? And the swing of the driver? Yikes.
Certainly the easiest path would be to raise B+ and settle for A1, or figure that distortion at 35+ watts out would be less noticeable if A2 was broached. Maybe not as challenging or fun to design though? PSUD says a 2uF lead cap would yield 830V and a 4uF would give 960V. The main filter caps I have are 20uF at 660VAC. I'm guessing they would take a kilovolt of DC okay.
Decisions, decisions. Dang
Michael
Hey Nic,
You have a wealth of good valve suggestions. Since I like the 12B4 a lot I especially like the 6CK4. Will have to get a few of those to try in a standalone. The question is, for me because I have no idea, can any valve drive a 2.25:2 interstage into grid current? Or does the stepdown need to be larger? And the swing of the driver? Yikes.
Certainly the easiest path would be to raise B+ and settle for A1, or figure that distortion at 35+ watts out would be less noticeable if A2 was broached. Maybe not as challenging or fun to design though? PSUD says a 2uF lead cap would yield 830V and a 4uF would give 960V. The main filter caps I have are 20uF at 660VAC. I'm guessing they would take a kilovolt of DC okay.
Decisions, decisions. Dang
Michael
Re: valve suggestions
...hmmm , maybe A2 is a bit too much with the IT . Thinking about it , a doubling of power is only 3dB , is it really worth it for all the increase in distortion / driver ? Maybe drop down to the 6.6k tapping at >800V/90mA will squeeze out what you require . Perhaps for A2 at low volts some really beefy direct coupled cathode followers are in order . EL38/6CN6 triode connected has Pdiss of 25w and approx 13mA/V transconductance , high mu of 16 as triodes so not too much gain would be lost (mu/mu+1) . That should do it ! Your caps should be good for 660x1.414 volts DC , I'd factor in mains fluctuations too , maybe ok as the reservoirs but not before the choke
ps For class A2 stuff have you looked at Dhaen's site ? Another good one is Sakuma : 845 driving 845 :-nuts
cheers
316a
audiobot said:
...hmmm , maybe A2 is a bit too much with the IT . Thinking about it , a doubling of power is only 3dB , is it really worth it for all the increase in distortion / driver ? Maybe drop down to the 6.6k tapping at >800V/90mA will squeeze out what you require . Perhaps for A2 at low volts some really beefy direct coupled cathode followers are in order . EL38/6CN6 triode connected has Pdiss of 25w and approx 13mA/V transconductance , high mu of 16 as triodes so not too much gain would be lost (mu/mu+1) . That should do it ! Your caps should be good for 660x1.414 volts DC , I'd factor in mains fluctuations too , maybe ok as the reservoirs but not before the choke
ps For class A2 stuff have you looked at Dhaen's site ? Another good one is Sakuma : 845 driving 845 :-nuts
cheers
316a
hi
If so, model of 12b4a are definitely wrong or the interstage
trafo has to be modelled as a nonlinear core.
At 1% distortion level, I obtain 135Vrms and 200Vrms (V(a,b)) with 10k and 60k respectively loading the secondary.
max 12b4a grid not positive (with respect to cathode) but near 0 volts.
circuit like the following
Federico
If so, model of 12b4a are definitely wrong or the interstage
trafo has to be modelled as a nonlinear core.
At 1% distortion level, I obtain 135Vrms and 200Vrms (V(a,b)) with 10k and 60k respectively loading the secondary.
max 12b4a grid not positive (with respect to cathode) but near 0 volts.
circuit like the following
Federico
Attachments
agreed
Hi Federico,
I agree, in that I think it should work. Although I am much more interested in the 0.1% distortion level than the 1%. Any idea of what I could be doing wrong, if such were to be the case? Also, what effect do you find from adding the 80k resistance across the primary? I have not really seen much change caused by varying it from 100k to 30k or removing it, except the output level drops a little bit as it's value decreases.
I will try the measurements again this morning and collect the distortion data for the two cases.
Later,
Michael
Hi Federico,
I agree, in that I think it should work. Although I am much more interested in the 0.1% distortion level than the 1%. Any idea of what I could be doing wrong, if such were to be the case? Also, what effect do you find from adding the 80k resistance across the primary? I have not really seen much change caused by varying it from 100k to 30k or removing it, except the output level drops a little bit as it's value decreases.
I will try the measurements again this morning and collect the distortion data for the two cases.
Later,
Michael
yes
divide by two my values so
100Vrms and 68 Vrms (60k and 10k)
so the difference is only with 10k load.
but putting the 100u cap between CT and ground decreases
dist. so 100Vrms , 0.2% dist with 60k. This does not
work if load is 10k: dist remains .9%.
What do you mean with
"great distortion when grid is driven positive" ?
How much dist?
Bye
Federico
divide by two my values so
100Vrms and 68 Vrms (60k and 10k)
so the difference is only with 10k load.
but putting the 100u cap between CT and ground decreases
dist. so 100Vrms , 0.2% dist with 60k. This does not
work if load is 10k: dist remains .9%.
What do you mean with
"great distortion when grid is driven positive" ?
How much dist?
Bye
Federico
first, I always load a stage with something.
Then, I find differences between loading before the primary or after the secondary of the trafo ( a part from the different reflected resistance).
If I load the trafo before the primary the it is almost not loaded and I find better freq. response and less resonance problems.
So, usually I put a main res between anode (i.e. the 80k res)
and a bigger one ( that practically does not load the trafo) at the secondary (e.g. 470k).
Federico
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.