Francois G - Thank you for taking the time to write such a comprehensive reply. I have been in touch with Dynaco for cost of Dynaco Z565 output transformers and they replied very quickly.
I see that they do a 240v kit for the SC35 which includes the Dynaco Z565.
Looking at the power supply is see that there is
PB-028-1 POWER XFMR
Description
240 vac 50/60 hz primary, secondary 330-0-330 vac @ 180 ma (dc) rectified, plus (2) 6.3 volt taps @2.5a. For mounting and wiring details, refer to dimensional/electrical schematic. Exact drop-in replacement for Dynakit ST-35 and SCA-35 amplifiers
PB-028-1 POWER XFMR - Dynakit Parts
Another option may be the Edcore Transformers which I can purchase from Don Audio in Spain i.e. EU
Any thoughts?
Thanks again
Lee
I see that they do a 240v kit for the SC35 which includes the Dynaco Z565.
Looking at the power supply is see that there is
PB-028-1 POWER XFMR
Description
240 vac 50/60 hz primary, secondary 330-0-330 vac @ 180 ma (dc) rectified, plus (2) 6.3 volt taps @2.5a. For mounting and wiring details, refer to dimensional/electrical schematic. Exact drop-in replacement for Dynakit ST-35 and SCA-35 amplifiers
PB-028-1 POWER XFMR - Dynakit Parts
Another option may be the Edcore Transformers which I can purchase from Don Audio in Spain i.e. EU
Any thoughts?
Thanks again
Lee
Dear Francois,
thanks for your head-up on the matter.
Based on gingertubes' first post on BH thread ( EL84 Amp - Baby Huey ), that configuration came from Yves Monmagnon's ECL86 amp ( Push Pull ECL86/6GW8 ).
From the simulations I've done, that configuration 220 kOhm + 47 kOhm with a bias of 600 uA gives very low THD with very low phases. If you divert from that, everything goes worse (higher THD and/or higher order harmonics content). At least from the simulations I've done.
I would like to understand more about the effect of the phases of the harmonics, but from Pass discussions, it seems the closer to zero, the better.
A good point of the EL84 and 12ax7 combo is that the former requires small signals to get full power, so the shunt feedback can be increased (and therefore the AC loadline of the phase inverter can become steeper) with small impact on sound. But for sure the shunt feedback is a good way to let the phase inverter and power amp reach their own limits together.
I'm trying to adapt it to other power amps (the "BH goes into puberty" thread), and by now similar results can be achieved with 6550s having plates at 600V and screens with a tertiary winding at 300V, and the shunt feedback taken from screens (and scaled up, because you'll have less signal on screens compared to plates). The lower the bias voltage needed by the output tube, the wider the swing demanded to the 12ax7, the worst the THD of the driver stage. More than acceptable for an instrument amplifier (that's why I'm designing the bass amp with a sextet of GU50s), but not for an Hi-Fi amp.
An interesting result obtained from the simulations (but I've still no clue on how this will fit into reality), is that there are two sweet spots for the PI: one more obvious biasing right in the middle of the available voltage, and another one biasing at a way lower voltage, with the PI working at higher currents (around 1 mA) giving an unsymmetrical output, but with low distortion on the positive swing.
I will post some results, that for sure will need more refinements from people with more experience in Hi-Fi than me.
Last edited:
Great to see others taking some ownership/leadership for this thread.
Also great to see some guys subjecting the design to modelling.
How good is modelling? The simple answer is that it is as good as the models used. In general the models for tubes do not handle the limits (that is tube cutoff or tube saturation) well. Particularly with saturation where you also have elevated grid current
So how did I handle the 12AX7 operating points in the original design.
We had (as usual) conflicting requirements.
1) For maximum output tube drive we need to be biased for equal positive and negative signal swings at the diff amp anode. That means anodes at 1/2 to 2/3 B+ (I lean toe=ward the 2/3rds).
2) Given The fact that the anode load resistors are "anti bootstrapped" By the BH scheme make them as high as possible.
3) The shunt feedback is divided by the rp looking down into the anode of the tube. At typical anode currents this is likely to be around 60K.
From the above we lean toward:
Low anode Current => High rp (to maximise the shunt feedback linearization of the output tubes).
So I plucked 270K anode loads from my educated butt.
The 16K feedback resistor was fine tuned by extensive listening to all sorts of music.
My speakers are a simple Cap to the tweaters, with a benign impedance vs frequency so I needed less feedback than others may require.
If your music is coloured by the Speaker Impedance vs Frequency curve the push that 16K up to 18K or 22K. I always found the 27K was too much. For your info: I also listened to it with 39K - it sounded just like a reasonable solid state amp, back to the drawing board.
Is tuning that cross connect, shunt feedback set, resistor that hard?
No, the sweat spot when simply listening, is quite apparent in that the 2 sound sources (your speakers) suddenly fade out and you are left with just a wall of music.
At this point, regardless of all the fancy math and modelling we have done (which may or may not have supported our design decisions) , the amp is finished.
To put it in the "Audio Phool" terminology we are looking for PRAT, Pace, Rhythm and Attack.
When you find this sweat spot then screw down the covers and leave kit alone. Thankfully the sweat spot range (for that "16K" resistor) is reasonably wide.
So while I am enjoying the modelling results, I'm reminded that this design ended up where it is with only the single best test equipment, 1 pair of ears.
Cheers
Ian
.
Also great to see some guys subjecting the design to modelling.
How good is modelling? The simple answer is that it is as good as the models used. In general the models for tubes do not handle the limits (that is tube cutoff or tube saturation) well. Particularly with saturation where you also have elevated grid current
So how did I handle the 12AX7 operating points in the original design.
We had (as usual) conflicting requirements.
1) For maximum output tube drive we need to be biased for equal positive and negative signal swings at the diff amp anode. That means anodes at 1/2 to 2/3 B+ (I lean toe=ward the 2/3rds).
2) Given The fact that the anode load resistors are "anti bootstrapped" By the BH scheme make them as high as possible.
3) The shunt feedback is divided by the rp looking down into the anode of the tube. At typical anode currents this is likely to be around 60K.
From the above we lean toward:
Low anode Current => High rp (to maximise the shunt feedback linearization of the output tubes).
So I plucked 270K anode loads from my educated butt.
The 16K feedback resistor was fine tuned by extensive listening to all sorts of music.
My speakers are a simple Cap to the tweaters, with a benign impedance vs frequency so I needed less feedback than others may require.
If your music is coloured by the Speaker Impedance vs Frequency curve the push that 16K up to 18K or 22K. I always found the 27K was too much. For your info: I also listened to it with 39K - it sounded just like a reasonable solid state amp, back to the drawing board.
Is tuning that cross connect, shunt feedback set, resistor that hard?
No, the sweat spot when simply listening, is quite apparent in that the 2 sound sources (your speakers) suddenly fade out and you are left with just a wall of music.
At this point, regardless of all the fancy math and modelling we have done (which may or may not have supported our design decisions) , the amp is finished.
To put it in the "Audio Phool" terminology we are looking for PRAT, Pace, Rhythm and Attack.
When you find this sweat spot then screw down the covers and leave kit alone. Thankfully the sweat spot range (for that "16K" resistor) is reasonably wide.
So while I am enjoying the modelling results, I'm reminded that this design ended up where it is with only the single best test equipment, 1 pair of ears.
Cheers
Ian
.
Last edited:
i like BH
I liked that !
Speakers, signal source, are more influential than small differences in laboratory measurements,
I am looking for WOW! In my old ears, like when I put in a vulgar amplifier designed in the 60s or 70s, Ei Yugoslavia valves that had certainty were false and turned out not to be.
Of course, I will test from 39K (first value in MK II) to 16K, only adjustment measures and what my ears think
Regards
Fernan
(I saw in a DAC written 0.00001% THD, in another 0.3%, can someone tell me which one sounds better, and why?)
I liked that !
Speakers, signal source, are more influential than small differences in laboratory measurements,
I am looking for WOW! In my old ears, like when I put in a vulgar amplifier designed in the 60s or 70s, Ei Yugoslavia valves that had certainty were false and turned out not to be.
Of course, I will test from 39K (first value in MK II) to 16K, only adjustment measures and what my ears think
Regards
Fernan
(I saw in a DAC written 0.00001% THD, in another 0.3%, can someone tell me which one sounds better, and why?)
Thanks Ian, so with an higher B+ you would keep the current low and raise the 220 kOhm and 47 kOhm (and therefore the 16k kOhm as well) values?
Just to make an example:
With 320 V I have 267 kOhm and 600 uA, so I drop 162 V.
If I use an EL34 at 450 V, would you go for something like 270-330 kOhm and 56-68 kOhm keeping the same current?
Thanks
Hello Lee,
While the Z565 clone Dynakit output transformer would be excellent for the Baby Huey (BH) the SCA35 power transformer is NOT ideal.
The BH PCB depends on fixed bias while the SCA35 runs with cathode bias which means that the SCA35 power transformer provides too high a B+, unless you fiddle with it to lower the supply voltage. The ideal BH EL84 power transformer provides about 230 Vac for B+ after rectification of around 300vdc, with 40-50 Vac bias tap and sufficient heater current to be rectified and dropped to 6.3 Vdc for the heaters. The first and last are requirements are relatively easy to satisfy, but you may have to buy a separate small 40-50 Vac transformer to provide the bias current and voltage. I have not checked the Edcore transformers but I would suggest you buy close to home if possible, due to the high cost of shipping transformers.
Good luck.
Last edited:
Ian, it is great to have you back! I was concerned that Covid-19 may have interfered with your newly earned retirement.
A question, not to upset your retirement state of mind:
Would a different high gain triode (think EC88) with higher gm have been better?
I would be very interested to hear about your listening observations of your BH EL34, compared to your favorite 6V6 implementation.
It is the output tubes which need to be higher gm.
The diffamp triodes need high rp and low anode currents, so EC88 rp of 4K8 will be too low. Ditto for ECC88/6DJ8 etc.
Something like an ECC807 would be great but "Rocking Horse Poo" availability, and what I have I'm saving for the Rogers Cadet which is my wake up clock radio amp driving some vintage Warfdales.
Bathroom renovation and gardening chores done so back to the workshop soon.
Cheers
Ian
The diffamp triodes need high rp and low anode currents, so EC88 rp of 4K8 will be too low. Ditto for ECC88/6DJ8 etc.
Something like an ECC807 would be great but "Rocking Horse Poo" availability, and what I have I'm saving for the Rogers Cadet which is my wake up clock radio amp driving some vintage Warfdales.
Bathroom renovation and gardening chores done so back to the workshop soon.
Cheers
Ian
Last edited:
I would like to ask a question about the effective rp of the EL84 after UL and shunt feedback are applied.
rp with shunt feedback is (from Shade pdf: http://www.mcmlv.org/Archive/TubeTheory/Schade 1938 Beam Power Tubes.pdf )
rp(r) = rp || [1 / (ns x gm)]
with ns = shunt feedback as 1 = 100%.
and rp with distributed load is (from Williamson and Walker)
rp(r) = rp / [ 1 + (nd x pentodemu / triodemu)]
with nd = percentage of turns as 1 = 100%.
So is the combination of the two:
{rp || [1 / (ns x gm)]} / [ 1 + (nd x pentodemu / triodemu)]
or it is needed to calculate it differently?
Thank you all in advance for the reply.
rp with shunt feedback is (from Shade pdf: http://www.mcmlv.org/Archive/TubeTheory/Schade 1938 Beam Power Tubes.pdf )
rp(r) = rp || [1 / (ns x gm)]
with ns = shunt feedback as 1 = 100%.
and rp with distributed load is (from Williamson and Walker)
rp(r) = rp / [ 1 + (nd x pentodemu / triodemu)]
with nd = percentage of turns as 1 = 100%.
So is the combination of the two:
{rp || [1 / (ns x gm)]} / [ 1 + (nd x pentodemu / triodemu)]
or it is needed to calculate it differently?
Thank you all in advance for the reply.
Bias/MOSFET supply
I'm looking at building a couple BH into a pair of existing monobloc chassis that have no source of 40 to 50VAC for the bias/MOSFET supply.
I found a Triad VPT48-1040 48VCT@1A toroid that looks suitable. Still not exactly cheap when shipped to Australia.
Is anyone aware of a small switcher, boost or buck converter module, a pair of which could be used to supply ~ ±65V?
They seem to max out at 35V.
Thanks!
I'm looking at building a couple BH into a pair of existing monobloc chassis that have no source of 40 to 50VAC for the bias/MOSFET supply.
I found a Triad VPT48-1040 48VCT@1A toroid that looks suitable. Still not exactly cheap when shipped to Australia.
Is anyone aware of a small switcher, boost or buck converter module, a pair of which could be used to supply ~ ±65V?
They seem to max out at 35V.
Thanks!
Here is an idea to consider. You do not need +-65 volts for an EL84 based BH. The rule of thumb is 3 x the - grid voltage on the power tube. For an EL84 this is usually in the -8 to-12v range . So +-35 would be fine. I recently completed a BH EL84 and actually use +20 and -36 for the bias supply without issues. There are a few resistors that may need to be changed for these voltages to be ideal, but +-65 is not needed. For an el34 version you need the higher voltages. Personally, I would stick with a linear supply and not use a switching supply for the bias voltages. There is a good chance that with a switching supply you could introduce HF noise or leakage current issues into the most sensitive part of the output stage.
Last edited:
Thanks for the suggestion. I'll play around with my simulation and get some starting points for the resistor values. It's a pretty cheap experiment if there are HF noise or leakage current issues.
Has anyone tried this amp with Z-565 output transformers and UL taps at 25%?
Or has anyone informations on how that transformer was done?
In Italy there's inmadout that produces good replicas of various transformers and we could organize a group buy for european users for both power transformer(s) and output transformers.
Or has anyone informations on how that transformer was done?
In Italy there's inmadout that produces good replicas of various transformers and we could organize a group buy for european users for both power transformer(s) and output transformers.
Great idea, Roberto. I would participate, however, I have a pair of unused Z-565 transformers on the shelf, as well as a pair of the Toroidy EL84s. The latter has 43% feedback, rather than 25%. My intent was to listen for and try to measure differences in performance to report back here. Unfortunately my hearing is failing, so my results may differ from yours.
Thanks Bas. My simulation suggests 10-15mA is indeed fine.
The lower voltage suggestion gives plenty of options. Small 24V and 30V 150mA transformers are available locally for under $10.
I'll do the same.
That would be great! I will keep this thread attentioned for your opinion on both. If you have infos on how those OTs are winded-up, I could ask a price for a group buy at inmadout.
Thanks in advance
Roberto, I have access to a simple Oscilloscope, what measurements should I take, since, unfortunately, I cant trust my failing ears.