I have seen schematics that connect one side of the secondary to circuit ground and others that don't😕 I have also built amps both ways and never noticed any difference. Is there a correct way to do this and what would be the benefits? Thanks🙂
The floating (not grounded) secondary sometimes gives better HF behaviour.... but dangerous!
Even at larger B+ (the swing on the primary is 2x B+) the primary-secondary breakthrough can cause electric shock!
Even at larger B+ (the swing on the primary is 2x B+) the primary-secondary breakthrough can cause electric shock!
Different applications have different reasons for design choices. Vintage public address systems often ran floating secondaries out to speakers, to allow continued operation should one cable accidently become grounded (single fault tolerance).
Floating the secondary winding may have some HF performance influence, as it could vary the stray capacitance levels, but that could only be a possible benefit for a valve amplifier without global negative feedback (as typical electrical signal feedback requires the secondary to connect to primary side circuitry).
Nowadays, there are safety related standards that don't allow floating voltage supplies if the voltage is sufficiently high - as occurs for higher powered valve amps feeding say 16 ohm speakers.
Floating the secondary winding may have some HF performance influence, as it could vary the stray capacitance levels, but that could only be a possible benefit for a valve amplifier without global negative feedback (as typical electrical signal feedback requires the secondary to connect to primary side circuitry).
Nowadays, there are safety related standards that don't allow floating voltage supplies if the voltage is sufficiently high - as occurs for higher powered valve amps feeding say 16 ohm speakers.
1. If there is a very large capacitance from the plate connection of the primary winding to the secondary winding, then grounding the secondary can cause high frequency roll off.
A simple test . . .
Short All the primary leads together.
Short All the secondary leads together.
Measure the primary to secondary capacitance (many handheld DMMs have a capacitance measurement).
Calculate the Capacitive Reactance of the measured primary to secondary capacitance (at 20kHz).
Capacitive Reactance, Xc = 1/(2 x pi x 20,000Hz x Capacitance)
Compare Xc versus the rated primary impedance, Z.
If Xc is not quite a bit more than Z, that is a bad transformer.
If that is the case, find a better output transformer, and replace the poorly designed output transformer.
Or, connect one lead of the secondary to ground, and live with the reduced high frequency bandwidth of the bad transformer.
Safety First!
Ground one connection of the secondary.
Prevent the "Surviving Spouse Syndrome"
2. Another safety issue:
I had a brand new output transformer.
The primary was shorted to the transformer laminations and end bells.
That would put B+ on the frame. Dangerous!
Even if a transformer is good new, it can go bad later, so be sure to ground the transformer laminations and end bells.
Safety First!
A simple test . . .
Short All the primary leads together.
Short All the secondary leads together.
Measure the primary to secondary capacitance (many handheld DMMs have a capacitance measurement).
Calculate the Capacitive Reactance of the measured primary to secondary capacitance (at 20kHz).
Capacitive Reactance, Xc = 1/(2 x pi x 20,000Hz x Capacitance)
Compare Xc versus the rated primary impedance, Z.
If Xc is not quite a bit more than Z, that is a bad transformer.
If that is the case, find a better output transformer, and replace the poorly designed output transformer.
Or, connect one lead of the secondary to ground, and live with the reduced high frequency bandwidth of the bad transformer.
Safety First!
Ground one connection of the secondary.
Prevent the "Surviving Spouse Syndrome"
2. Another safety issue:
I had a brand new output transformer.
The primary was shorted to the transformer laminations and end bells.
That would put B+ on the frame. Dangerous!
Even if a transformer is good new, it can go bad later, so be sure to ground the transformer laminations and end bells.
Safety First!
Last edited:
Show schematic or it didn´t happen.
We don´t know what "you saw"
Really? 😕
Unless your amp has NO negative feedback, which is not impossible but quite unusual, you need a ground reference, a.k.a. ground connection.
Follow the schematic.
Again, no ground connection no NFB.
You are talking about a non NFB amp?
If so, say it.
Verry unlikely and a gross failure, unrelated to what´s being asked (which assumes a normal working OT)
Last edited:
For a PP amp, the effective shunt capacitance from a plate to B+ includes a variety of capacitance paths including from the plate end to core and to secondary. For common amplifier setups with the secondary connected to circuit ground, the shunt capacitance is measured with core and secondary connected to B+ winding end.
The shunt capacitance is only a part of the story for HF performance, as the leakage inductance and the total symmetry/configuration/feedback of the OPT, and the amplifier/speaker impedances make up the rest of the HF story that identifies when the first and following resonances occur and to what Q level.
Just saying that making any conclusions from a capacitance test needs to be done with caution.
The shunt capacitance is only a part of the story for HF performance, as the leakage inductance and the total symmetry/configuration/feedback of the OPT, and the amplifier/speaker impedances make up the rest of the HF story that identifies when the first and following resonances occur and to what Q level.
Just saying that making any conclusions from a capacitance test needs to be done with caution.
trobbins,
Since my first test does not work in all situations, then here is another one (no single test method works in all situations).
Open up the Global feedback loop, if there is one.
Get a DMM that goes to 20kHz.
Check the level at 20kHz with an 8 Ohm load across Common and 8 taps, and with Common grounded.
Then check the level at 20kHz with an 8 Ohm load across Common and 8 taps, and with the secondary floating.
Compare the 2 readings.
If it drops off, you know something for sure, the capacitance is having an effect.
How much, you be the judge.
Or, just measure the effect with your ear,
Oh, my ears do not go to 20kHz anymore.
Then you may receive a possible shock if you forget to re-ground the secondary (and to make the Global negative feedback work again).
If things like leakage reactance and winding to laminations and capacitance to end bells are causing problems at 20kHz . . .
Then you might want to consider replacing the output transformer with a better one.
Good ones usually have those problems at frequencies higher than 20kHz.
Just my opinion
Since my first test does not work in all situations, then here is another one (no single test method works in all situations).
Open up the Global feedback loop, if there is one.
Get a DMM that goes to 20kHz.
Check the level at 20kHz with an 8 Ohm load across Common and 8 taps, and with Common grounded.
Then check the level at 20kHz with an 8 Ohm load across Common and 8 taps, and with the secondary floating.
Compare the 2 readings.
If it drops off, you know something for sure, the capacitance is having an effect.
How much, you be the judge.
Or, just measure the effect with your ear,
Oh, my ears do not go to 20kHz anymore.
Then you may receive a possible shock if you forget to re-ground the secondary (and to make the Global negative feedback work again).
If things like leakage reactance and winding to laminations and capacitance to end bells are causing problems at 20kHz . . .
Then you might want to consider replacing the output transformer with a better one.
Good ones usually have those problems at frequencies higher than 20kHz.
Just my opinion
Last edited:
My concern is if that decision scenario is generally applicable (ie. a lowered capacitance due to a floating secondary causes a response change and so indicates a poor transformer).
If the amp has GNF then it has to work with the secondary grounded, so a test that ungrounds the secondary and then looks for differences down at the 20kHz end may (I caution) not be well aligned with performance when GNF is applied and resonance response is significantly above 20kHz.
Lee [1957] indicates the likely change in frequency response for a loaded OPT when shunt capacitance is lowered. However, the change in response depends on the driving resistance of the plate circuitry (eg. an amp that uses triode compared to pentode mode may have a substantial influence), and on the ratio of the leakage inductance impedance to the shunt capacitance impedance (which is being changed by the test conditions).
I have only measured OPT's for primary shunt capacitance with the standard test configuration (secondary grounded at one point), and only for some hi-fi quality OPT's for a Williamson amp, so don't have any good figures for what drop in capacitance to expect from the change in test setup. I am slowly gathering momentum to have another go at a Williamson amp, so will add that my list of tests.
I note that even for a hi-fi OPT in an amp with GNF disconnected, the drop in response at 20kHz can be quite noticeable - even Williamson measured about 2dB down.
If the amp has GNF then it has to work with the secondary grounded, so a test that ungrounds the secondary and then looks for differences down at the 20kHz end may (I caution) not be well aligned with performance when GNF is applied and resonance response is significantly above 20kHz.
Lee [1957] indicates the likely change in frequency response for a loaded OPT when shunt capacitance is lowered. However, the change in response depends on the driving resistance of the plate circuitry (eg. an amp that uses triode compared to pentode mode may have a substantial influence), and on the ratio of the leakage inductance impedance to the shunt capacitance impedance (which is being changed by the test conditions).
I have only measured OPT's for primary shunt capacitance with the standard test configuration (secondary grounded at one point), and only for some hi-fi quality OPT's for a Williamson amp, so don't have any good figures for what drop in capacitance to expect from the change in test setup. I am slowly gathering momentum to have another go at a Williamson amp, so will add that my list of tests.
I note that even for a hi-fi OPT in an amp with GNF disconnected, the drop in response at 20kHz can be quite noticeable - even Williamson measured about 2dB down.
Last edited:
Without negative feedback from either the primary, or from the secondary . . .
when the high plate resistance, rp, of a Pentode, or of a Beam Power is driving lots of capacitance, the high frequency response will be attenuated.
That can hold true for both push pull, and for single ended output stages.
Negative feedback can raise the high frequency response to back to flat.
How much negative feedback is needed is variable, and some factors are:
The leakage reactance of the output transformer
The various capacitances of the output transformer
The impedance versus frequency of the loudspeaker load
The degree of difficulty of adjusting negative feedback is dependent on the above factors, and also includes the rest of the amplifier circuitry, gain, phase, and frequency response, etc.
Did one of the first Williamson amplifiers use Triode Wired Beam Power tubes?
That certainly would have lowered the plate resistance, rp, that drove the transformer.
Most of engineering involves making tradeoffs, which is one of the reasons that certain
circuits are still being used in modern amplifiers, Williamson, Mullard, McIntosh Unity Coupled transformer, etc.
All of these work very well, depending on how well they are executed, including the selection of the output transformer.
when the high plate resistance, rp, of a Pentode, or of a Beam Power is driving lots of capacitance, the high frequency response will be attenuated.
That can hold true for both push pull, and for single ended output stages.
Negative feedback can raise the high frequency response to back to flat.
How much negative feedback is needed is variable, and some factors are:
The leakage reactance of the output transformer
The various capacitances of the output transformer
The impedance versus frequency of the loudspeaker load
The degree of difficulty of adjusting negative feedback is dependent on the above factors, and also includes the rest of the amplifier circuitry, gain, phase, and frequency response, etc.
Did one of the first Williamson amplifiers use Triode Wired Beam Power tubes?
That certainly would have lowered the plate resistance, rp, that drove the transformer.
Most of engineering involves making tradeoffs, which is one of the reasons that certain
circuits are still being used in modern amplifiers, Williamson, Mullard, McIntosh Unity Coupled transformer, etc.
All of these work very well, depending on how well they are executed, including the selection of the output transformer.
Yeh, Williamson triode strapped the KT66 - the nominal plate resistance, rp, dropped from 22.5k to 1k6. Off-shore in Australia the 807 was immediately substituted (as it was plentiful and cheap), as the triode strapped rp was only a titch higher at 1k9 and it worked at the nominal B+ (and the 807 screen voltage ratings were subsequently tested and then rerated for 400V).
Partridge's first commercial OPT offering, the WWFB, appears to have a nominal PP shunt capacitance of 500pF and 15mH leakage for a 60kHz first resonance, although I haven't yet tried to measure that on my one and only sample. Within 2 years Partridge offered the CFB range that pushed the first resonance up to 65kHz by lowering leakage to 10mH, but the compromise was a higher shunt capacitance of 600pF. Sadly I don't have a CFB sample. Interestingly, Dixon in the 1953 NRL report 4136 didn't get a better response from the CFB.
Partridge's first commercial OPT offering, the WWFB, appears to have a nominal PP shunt capacitance of 500pF and 15mH leakage for a 60kHz first resonance, although I haven't yet tried to measure that on my one and only sample. Within 2 years Partridge offered the CFB range that pushed the first resonance up to 65kHz by lowering leakage to 10mH, but the compromise was a higher shunt capacitance of 600pF. Sadly I don't have a CFB sample. Interestingly, Dixon in the 1953 NRL report 4136 didn't get a better response from the CFB.
Last edited:
There is a credible safety issue in floating an output transformers' secondary and probably should be condemned for that reason. Any reasonable risk to your children or pets is too much risk. Please just don't do it.
YOS,
Chris
YOS,
Chris
I remember we've had a similar discussion not too long ago, and my arguments remain the same: At least in their MC-3500 and MI-350 amplifiers McIntosh allowed the user to ground or not to ground the speaker windings. NFB was taken form one of the pentafilarly wound primary windings. In addition, I've seen output trannys/amplifier designs with a floating secondary winding just for safety reasons in TV sets and table radios with live chassis and external speaker sockets.
Hence, there's no general and mandatory to do or not to do with it. It depends.
Best regards!
Hence, there's no general and mandatory to do or not to do with it. It depends.
Best regards!
Last edited:
My amps have no global nfb and the difference grounding makes to the sound is clearly obvious. That settles the issue for me. Perhaps some transformer designs are less sensitive to this.
NFB
I have never built anything with NFB and have never attempted a PP design. Nor have any ever use what I would consider high voltage, > 500 volts. So if this is a matter of safety it would have be a crappy output transformer that could not withstand twice the B+. I would not swear by it but I am confident that most transformers I have used have a fairly high hi-pot value😀
I have never built anything with NFB and have never attempted a PP design. Nor have any ever use what I would consider high voltage, > 500 volts. So if this is a matter of safety it would have be a crappy output transformer that could not withstand twice the B+. I would not swear by it but I am confident that most transformers I have used have a fairly high hi-pot value😀
Current safety standards require that the secondary be grounded or double insulated from the primary working voltage (DC voltage plus peak of signal - essential twice B+). Double insulation for 600V would require testing primary to secondary with 5000V AC and that internal creepage (spacing between primary and secondary windings across insulation) of at least 4 mm. Good luck making a decent output transformers under those constraints!
All commercial high plate tube amplifier should have their secondary grounded as a standard safety rule. A floating secondary is dangerous in case for some reason plate voltage appears across the secondary terminals. Such reason could be insulation breakdown for example.
When DIY, you can float the secondary at your own personal risk.
A competent transformer designer has the choice to shift the overall transformer capacitance in the following ways
-towards primary to primary layers, giving better HF performance at grounded secondary and the difference between grounded vs floating will be smallest.
-towards primary to secondary layers, resulting in the poorest HF performance with a grounded secondary, but much better one with floating secondary.
-balance between the two.
When DIY, you can float the secondary at your own personal risk.
A competent transformer designer has the choice to shift the overall transformer capacitance in the following ways
-towards primary to primary layers, giving better HF performance at grounded secondary and the difference between grounded vs floating will be smallest.
-towards primary to secondary layers, resulting in the poorest HF performance with a grounded secondary, but much better one with floating secondary.
-balance between the two.
Last edited:
Nobody in this thread said which sounded better . . . grounded, or floating.
They just said it sounded different / measured different.
But . . . if you have real good high frequency hearing, and your loudspeakers have a nasty amplitude rise at high frequencies, then I bet the grounded secondary sounds better (as in grounded, with falling high frequency amplitudes, to undo the aggressive loudspeaker high frequency response).
Hi Fi / Stereo music playback systems are just that . . . a system.
I hope all your systems are Positive Synergy, not Negative Synergy.
Whatever you do for DIY, you may get away with violating all the safety rules.
But if anybody in your home gets hurt, and the Homeowner Insurance Company finds out about the violation, they will not pay for the injury.
How about life insurance? Might sue the builder.
Then, be sure to bring the amplifier to your friends for a listen.
Transformer insulation often degrades over time, temperature, and humidity.
Old Hot Chassis TVs with vacuum tubes, and output transformers had 4 characteristics,
1. No access to the chassis, they used a power mains automatic disconnect as soon as you pulled the back off.
2. There was no access to the speaker.
3. There was no access to the output transformer secondary.
4. The back cover had a warning statement: Danger, do not remove this panel. Access is for qualified service personnel only.
Different countries have different safety standards, and different Power Mains details.
I recommend that you meet all those standards, for the country you live in.
I ran Hi-Pot tests for 2 years when I was the technician for Spectrum Analyzer Power Supplies. Up to 55 supplies went out the door every week.
There were Lots of Hi-Pot failures, they were repaired or trashed, depending on the cause.
Your Mileage May Vary
____ Ethnic ____ Roulette with a revolver is a very bad game. Don't play that.
They just said it sounded different / measured different.
But . . . if you have real good high frequency hearing, and your loudspeakers have a nasty amplitude rise at high frequencies, then I bet the grounded secondary sounds better (as in grounded, with falling high frequency amplitudes, to undo the aggressive loudspeaker high frequency response).
Hi Fi / Stereo music playback systems are just that . . . a system.
I hope all your systems are Positive Synergy, not Negative Synergy.
Whatever you do for DIY, you may get away with violating all the safety rules.
But if anybody in your home gets hurt, and the Homeowner Insurance Company finds out about the violation, they will not pay for the injury.
How about life insurance? Might sue the builder.
Then, be sure to bring the amplifier to your friends for a listen.
Transformer insulation often degrades over time, temperature, and humidity.
Old Hot Chassis TVs with vacuum tubes, and output transformers had 4 characteristics,
1. No access to the chassis, they used a power mains automatic disconnect as soon as you pulled the back off.
2. There was no access to the speaker.
3. There was no access to the output transformer secondary.
4. The back cover had a warning statement: Danger, do not remove this panel. Access is for qualified service personnel only.
Different countries have different safety standards, and different Power Mains details.
I recommend that you meet all those standards, for the country you live in.
I ran Hi-Pot tests for 2 years when I was the technician for Spectrum Analyzer Power Supplies. Up to 55 supplies went out the door every week.
There were Lots of Hi-Pot failures, they were repaired or trashed, depending on the cause.
Your Mileage May Vary
____ Ethnic ____ Roulette with a revolver is a very bad game. Don't play that.
Remember that too much capacitive roll-off, which might occur more in cases of grounded secondary, will increase reflected load vs frequency, thus distortion in the HF region.
The Harmonic distortion of 10kHz is at 20kHz, 30kHz, 40kHz.
Yes, I know, those 10kHz tones are going to inter-modulate with with other tones too.
Do you have good hearing at 20kHz?
How good are your loudspeakers at 20kHz?
Are there dogs in your house?
Worry, Worry, Worry.
If your transformer is struggling at 10kHz, replace it with a better one.
Then you can stop worrying.
Yes, I know, those 10kHz tones are going to inter-modulate with with other tones too.
Do you have good hearing at 20kHz?
How good are your loudspeakers at 20kHz?
Are there dogs in your house?
Worry, Worry, Worry.
If your transformer is struggling at 10kHz, replace it with a better one.
Then you can stop worrying.
A small example. If we take a triode with 1k plate resistance and decide to use a 5k transformer primary load. Let's say the transformer has 4nF of overall capacitance, which gives an unloaded transformer a -3dB roll-off at 40kHz and a loaded one 48kHz. Not so bad at first sight.
Now 4nF of capacitance has a reactance of 1k at 40kHz and 5k and 8kHz.
Overall load seen by the tube becomes 2.5k at 8kHz, because you get both 5k loads (capacitive and real load in parralel) Think about that.
Now 4nF of capacitance has a reactance of 1k at 40kHz and 5k and 8kHz.
Overall load seen by the tube becomes 2.5k at 8kHz, because you get both 5k loads (capacitive and real load in parralel) Think about that.