I write many things about the transformer quality in my schematics, the primary inductance and the coupling factor. Here's some "Spice" measurements I made on the "global feedback version" with differents transformer parameters ...
In the next days, I will try the "global feedback version" with a lower quality transformer, the 37$ Edcor GXSE15-8-1,7K ... It is not a 1,25K like in the schematic but it is the closest. They don't said much about it and I don't know if it can stand the 110ma of the two parallel 6S19P ... It is only a test to see what the results can be with this cheaper transformer !
Yesterday, I look at about 70 russian triodes datasheets to see if some can be used instead of the expensive 6DJ8 and I think I find one but I have to make is Spice model and try it in my circuit ... I will be back about that ...
Alain.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
An externally hosted image should be here but it was not working when we last tested it.
In the next days, I will try the "global feedback version" with a lower quality transformer, the 37$ Edcor GXSE15-8-1,7K ... It is not a 1,25K like in the schematic but it is the closest. They don't said much about it and I don't know if it can stand the 110ma of the two parallel 6S19P ... It is only a test to see what the results can be with this cheaper transformer !
Yesterday, I look at about 70 russian triodes datasheets to see if some can be used instead of the expensive 6DJ8 and I think I find one but I have to make is Spice model and try it in my circuit ... I will be back about that ...
Alain.
Thank you !
Doing that is a pleasure for me and that's how I learn always more about tubes electronic and it can help other DIYers to design their own circuits.
I will probably test that circuit on a "breadboard" one of these days but I may not build it really on a "deluxe" steel frame with good parts and all the "tralala" because it is too expensive for my tiny budget ... But I have two others "very special" circuits idea using the 6S19P in mind, something I never see before on Internet ... But it is "top secret" until I test them carefully and put them on my personnal web site ...
Currently you have triode loaded on low resistance of parallel feedback getting significant distortions caused by variable Rp, i.e. feedback factor is defined by Rp of driving triode and Rfb /A. where A is voltage amplification factor of the power stage.
If you remove diodes from casthode and put there resistor, or even better use pentode to drive such low resistance, results will be even better: big resistance, almost current source, loaded on low input resistance of the power stage with parallel feedback.
All this was discussed many times here.
If you remove diodes from casthode and put there resistor, or even better use pentode to drive such low resistance, results will be even better: big resistance, almost current source, loaded on low input resistance of the power stage with parallel feedback.
All this was discussed many times here.
Yes, I am working on all that today ... Feedback change many things ...
The dynamic plate resistance have a lot to do with frequency response. I am trying to find the best way to evaluate the real transformer response with the simulator, finding is exact coupling factor from the specifications, EX: -1dB at 18Khz ...
I will also try to figure out if my SIMetrix simulator really calculate the output transformers the right way (current variation) and not like power transformers (voltage variation), this is very different ...
Alain.
The big problem using the "ideal transformer model" of my simulator, a "non saturating device", is I have to enter the coupling factor (K) and I don't know how to measure it or calculate it ...
The formula K = M / √ ( L1 x L2 ) suppose to give it but M is the "mutual inductance" and I don't know how to measure it or calculate it ...
L1 is often given for the output transformers primary but not L2, I suppose it must be about proportionnal to the turn ratio ... If L1 is 53H and the ratio is 12,5 ( for a 1250 to 8 ohms transformer ) the L2 inductance should not be very far from 53 / 12,5 = 4,24H ... Probably, M is related to the difference between the turn ratio and the inductance ratio ?????
K have to be less than 1 so if √ ( 53 x 4,24 ) = 14,99066376, M must be lower than that ... For example, if M = 14,9 , K will be 0,993951985 ... Those calculations have to be very precise or the "real" high frequency response can be very different from the "calculated" one ...
Anybody have an idea about all that ??????
Alain.
The formula K = M / √ ( L1 x L2 ) suppose to give it but M is the "mutual inductance" and I don't know how to measure it or calculate it ...
L1 is often given for the output transformers primary but not L2, I suppose it must be about proportionnal to the turn ratio ... If L1 is 53H and the ratio is 12,5 ( for a 1250 to 8 ohms transformer ) the L2 inductance should not be very far from 53 / 12,5 = 4,24H ... Probably, M is related to the difference between the turn ratio and the inductance ratio ?????
K have to be less than 1 so if √ ( 53 x 4,24 ) = 14,99066376, M must be lower than that ... For example, if M = 14,9 , K will be 0,993951985 ... Those calculations have to be very precise or the "real" high frequency response can be very different from the "calculated" one ...
Anybody have an idea about all that ??????
Alain.
I found this about Mutual inductance and coupling factor ...
I also found in my notes a trick to measure the coupling factor with the secondary open or shorted ...
L1 is the primary inductance with the secondary open.
LS is the primary inductance with the secondary shorted.
K = √ ( 1 - LS / L1 )
I try it once but this is not very precise and you need the physical transformer on hands to measure it ...
What I really like to know is a method to properly evaluate the coupling factor just with the others transformer specifications with the help of the simulator without having to buy the transformer first ...
I wonder why any output transformers factorys never give that important parameter or the core material exact type ... They seem to said "just buy them and don't ask any questions" ...
Alain.
I also found in my notes a trick to measure the coupling factor with the secondary open or shorted ...
L1 is the primary inductance with the secondary open.
LS is the primary inductance with the secondary shorted.
K = √ ( 1 - LS / L1 )
I try it once but this is not very precise and you need the physical transformer on hands to measure it ...
What I really like to know is a method to properly evaluate the coupling factor just with the others transformer specifications with the help of the simulator without having to buy the transformer first ...
I wonder why any output transformers factorys never give that important parameter or the core material exact type ... They seem to said "just buy them and don't ask any questions" ...
Alain.
Hi Alan when I was start reading this topic, spot that value for the Ck of output tube maybe should be lower than on the original sch, but You lower it on the next page...
About devices with hysteresis and core issues, also with reactive nature there is no
absolute model for audio simulations, that is not clear about it. maybe the best results are obtaining to avoid core issues and to try with simple equivalent circuit
like L(prim ind.) C(total capac.) R(load) in paralel with one Rdc serial circuit for the start
and monitor dynamic load line elipse shape
About devices with hysteresis and core issues, also with reactive nature there is no
absolute model for audio simulations, that is not clear about it. maybe the best results are obtaining to avoid core issues and to try with simple equivalent circuit
like L(prim ind.) C(total capac.) R(load) in paralel with one Rdc serial circuit for the start
and monitor dynamic load line elipse shape
My circuit evolve with time, I will also try it with a "low cost" transformer, also other driver and preamp tubes to see the differences ... Like I write before the problem I have with the "ideal" transformer model is I have to enter the "coupling factor" or the result mean nothing at frequency over 1Khz.
The primary inductance is closely related to the lower frequency response because a coil is opposed to current change and is phase is 90° later than the voltage one, so the real power is lower and the waveshapes get disformed and the distortion can raise very high ...
The coupling factor is closely related to the higher frequency response and hysteresis, the windings interleave and core material are very important to get low hysteresis, that's why the best output transformer use "oriented grain silicon steel" core material. But like you will see in my next post, everything depend on the "dynamic plate resistance" too.
Alain.
I made an interresting simulation experience today, I figure out the "dynamic plate resistance" (Rp or Ra) is a very important factor in the frequency response. The triode 6S19P have a very very low Rp (566,21 ohms) and a pentode or "tetrode beampower tube" like the 35C5 have a high Rp (10000 ohms) and that make all the difference ... Some tube like the 6L6 have a much higher Rp (35000 ohms) and it is very important to use a high primary inductance transformer for those or there will be not much bass ...
The two circuits results with exactly the same output transformer ...
Notice the SIMetrix frequency response curves are in "peak" volts and not RMS ... They also get more imprecise as the distortion raise at the lower and higher frequencys ...
Obviewsly, triodes need a high "coupling factor" transformer to avoid high frequency attenuation but the waveshape stay nice. They can be use with a low primary inductance without too much distortion at low frequency but it depend on the circuit too, fixed or cathode bias, feedback or no feedback, etc ...
The pentode or "tetrode beampower tube" have a much better response in high frequency with a low "coupling factor" but the hysterisis make more distortion ... However, if the primary inductance is too low, the results can be horribles ...
This experience tell me something, when the output transformers factorys claim their frequency response is from X Hz to Y Khz within 1dB, that is false publicity because they never say with what tube "dynamic plate resistance" it is ...
My next simulations will be using a Edcor "low cost" transformer with my two last fixed bias circuits, the one "without global feedback" and the one "with global feedback" ...
Alain.
An externally hosted image should be here but it was not working when we last tested it.
The two circuits results with exactly the same output transformer ...
An externally hosted image should be here but it was not working when we last tested it.
Notice the SIMetrix frequency response curves are in "peak" volts and not RMS ... They also get more imprecise as the distortion raise at the lower and higher frequencys ...
Obviewsly, triodes need a high "coupling factor" transformer to avoid high frequency attenuation but the waveshape stay nice. They can be use with a low primary inductance without too much distortion at low frequency but it depend on the circuit too, fixed or cathode bias, feedback or no feedback, etc ...
The pentode or "tetrode beampower tube" have a much better response in high frequency with a low "coupling factor" but the hysterisis make more distortion ... However, if the primary inductance is too low, the results can be horribles ...
This experience tell me something, when the output transformers factorys claim their frequency response is from X Hz to Y Khz within 1dB, that is false publicity because they never say with what tube "dynamic plate resistance" it is ...
My next simulations will be using a Edcor "low cost" transformer with my two last fixed bias circuits, the one "without global feedback" and the one "with global feedback" ...
Alain.
Last edited:
About using a "low cost" transformer with my better circuits
If you are a bit "cheap", "tire la cenne" or just poor like me, you may like to buy a 15W Edcor GX series at 37$ for an amplifier like that but ... Like I write before, I don't know the "coupling factor" of the Edcor output transformers and this series may not be suitable for a very low "dynamic plate resistance" triodes amplifier ...
This is at your own risk ...
Don't expect to have the same winding coupling quality as the 25W Edcor CXSE25-8-1,25K at 83,50$, a 30W "Classic" Hammond 1640SEA at 160$, a 25W Lundahl LL1627 at 260$, a LL1627AM at 450$ or a "custom hand made" Electra-Print at ???$
I use a 37$ 15W Edcor GXSE15-8-1,7K transformer in those circuits, because there is no 1,25K in that series and this is the closest, but I don't know if it can handle 110ma DC without too much DC core saturation ... Since it have a higher impedance, the power is about 80% lower than with the 1,25K and less if there is saturation ...
Because this transformer have a higher primary DC resistance, then a higher voltage drop across it, I have to raise the supply voltage and the R9 value a little bit to get the same voltages and currents.
This is the "no global feedback" version and the distortion is higher than with the CX series transformer, specially at 20Hz ... As you can see, if the "coupling factor" is less than 0,9992 the high frequency drop very fast and the sound can be really smooth ... Unless you use this amplifier for a small subwoofer or a bass guitar ...
Now the "global negative feedback" version :
As you can see, at 1Khz, the distorsion is lower than with the 1,25K transformer but higher at 20Hz ... But it is acceptable for the transformer price, the global negative feedback improve the high frequency response a little bit, a "average quality" 0,99 "coupling factor" can be acceptable if you have a treble boost on your preamp ... It is very easy to add one to this circuit with baxandall tone controls and a very good magnetic phono preamp but that is not my goal in this thread.
I have a hard time to design this one because there was a "peaks" problem at 20Khz and it made my simulation abort ... I have to add the C4 capacitor to correct the problem and depending on the "real" transformer coupling, C3 and C4 may have to be adjusted ... The problem was those high frequency "peaks" make the 6S19P grids slightly positives in their upper portion causing 0 bias clipping ...
I will not be back with those little transformers, I will now replace the 6DJ8 preamp with a good old 12AX7 to see if it is better for feedback because the cathode resistor will be much higher.
Alain.
If you are a bit "cheap", "tire la cenne" or just poor like me, you may like to buy a 15W Edcor GX series at 37$ for an amplifier like that but ... Like I write before, I don't know the "coupling factor" of the Edcor output transformers and this series may not be suitable for a very low "dynamic plate resistance" triodes amplifier ...
This is at your own risk ...
Don't expect to have the same winding coupling quality as the 25W Edcor CXSE25-8-1,25K at 83,50$, a 30W "Classic" Hammond 1640SEA at 160$, a 25W Lundahl LL1627 at 260$, a LL1627AM at 450$ or a "custom hand made" Electra-Print at ???$
I use a 37$ 15W Edcor GXSE15-8-1,7K transformer in those circuits, because there is no 1,25K in that series and this is the closest, but I don't know if it can handle 110ma DC without too much DC core saturation ... Since it have a higher impedance, the power is about 80% lower than with the 1,25K and less if there is saturation ...
Because this transformer have a higher primary DC resistance, then a higher voltage drop across it, I have to raise the supply voltage and the R9 value a little bit to get the same voltages and currents.
An externally hosted image should be here but it was not working when we last tested it.
This is the "no global feedback" version and the distortion is higher than with the CX series transformer, specially at 20Hz ... As you can see, if the "coupling factor" is less than 0,9992 the high frequency drop very fast and the sound can be really smooth ... Unless you use this amplifier for a small subwoofer or a bass guitar ...
Now the "global negative feedback" version :
An externally hosted image should be here but it was not working when we last tested it.
As you can see, at 1Khz, the distorsion is lower than with the 1,25K transformer but higher at 20Hz ... But it is acceptable for the transformer price, the global negative feedback improve the high frequency response a little bit, a "average quality" 0,99 "coupling factor" can be acceptable if you have a treble boost on your preamp ... It is very easy to add one to this circuit with baxandall tone controls and a very good magnetic phono preamp but that is not my goal in this thread.
I have a hard time to design this one because there was a "peaks" problem at 20Khz and it made my simulation abort ... I have to add the C4 capacitor to correct the problem and depending on the "real" transformer coupling, C3 and C4 may have to be adjusted ... The problem was those high frequency "peaks" make the 6S19P grids slightly positives in their upper portion causing 0 bias clipping ...
I will not be back with those little transformers, I will now replace the 6DJ8 preamp with a good old 12AX7 to see if it is better for feedback because the cathode resistor will be much higher.
Alain.
Wow ! I never ear about them before ... Giuseppe Bartolucci ...
They look really goods, the 52 model will do for sure ... 1250 to 4-8-16 ohms, 280ma DC, 18W, 16H, 41 ohms primary DCR, 20Hz to 78Khz within 3dB ... They don't mention the "coupling factor" but with a response up to 78Khz, it must be very high ...
But 300 Euros ... This is actually 401,61$ US ... For 2,50$ russian tubes ...
Not really a tiny budget price !
I will try driving the 6S19P with a 6AU6 pentode, they are very linear with a - 0,5V bias, but I need at least a 140V peak to peak swing ... I think I got about 50 of them in my "junkbox" and they cost about nothing on eBay when you buy a lot, the russian equivalent 6J4P is also very interesting. Those tubes can also give very good results triode connected ...
Have you tried to make your own output transformers ? It's really not very difficult, and it does not require winding machine - handmade is really better !
Gleb.
Wow ! I never ear about them before ... Giuseppe Bartolucci ...
They look really goods, the 52 model will do for sure ... 1250 to 4-8-16 ohms, 280ma DC, 18W, 16H, 41 ohms primary DCR, 20Hz to 78Khz within 3dB ... They don't mention the "coupling factor" but with a response up to 78Khz, it must be very high ...
But 300 Euros ... This is actually 401,61$ US ... For 2,50$ russian tubes ...
Not really a tiny budget price !
try to find good laminates and finish the windings and calcs by yot own work... might go cheeper a bit...
google for "Wasner" core laminates I think that is 15 EU per Kg
permeability of about 800 (classic is 400 acc.)
allowing the less number of turns for high prim ind.
and lower capacitance...
.
But prepare to loose a day just to cut and finish isolation papers...
Hi Alain,
As previously discussed many times before on this forum, do yourself a favor and use a pentode as a driver instead of a triode. This is the best way to implement a Schade feedback even though I tried it many time and did not see a real improvement. Some of the members on this forum sware by this ;-)
Rgds,
Eric
As previously discussed many times before on this forum, do yourself a favor and use a pentode as a driver instead of a triode. This is the best way to implement a Schade feedback even though I tried it many time and did not see a real improvement. Some of the members on this forum sware by this ;-)
Rgds,
Eric
Original Schade feedback (per his 1938 article) does not need pentode at all: unlike what is called by his name here, it was different kind of feedback applied in series with input signal, to the secondary winding of an interstage transformer.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.