The tech box off Fabrizio are inside my article on Audioreview magazine and , as told before, I sent on other thread the results and specs with explanation.
The test set is easy to build and with a minimum of equipments you can get lot of real information about trafo.
The simulation is not appropriate with a real trafo, too many difference are present from reality.
Walter
Walter
The test set is easy to build and with a minimum of equipments you can get lot of real information about trafo.
The simulation is not appropriate with a real trafo, too many difference are present from reality.
Walter
Walter
You have many posts, so not sure exactly which ones you are referring to... anyway, here are some that look relevant:
LITZ output trafo
LITZ output trafo
If these are not what you were referring to, you'd have to dig them out youself... 🙂
LITZ output trafo
LITZ output trafo
If these are not what you were referring to, you'd have to dig them out youself... 🙂
In the first figure I attached, the lower circuit of the two, in post 35 of this new thread. The VDV3035SE is rated 3486 to 4 Ohms, so in your setup as you described, you'd make R9 3486 Ohms as you can see in the picture. However, at this side, it is driven by 700 Ohms (the source resistance of the 300B) when playing in an amp, and that determines the HF behavior of the primary capacitance/leakage inductance parasitic lowpass filter. If you swap out the 3486 Ohms resistor for a 700 Ohms load resistor (change R9 to 700 Ohms), the resulting "backwards" response suddenly becomes almost perfectly matched with the "forward" response. I have adjusted the signal level to show that both curves are practically identical. The phase curves of the two sims are such a good match that they show as one curve, but there are really two in the picture n top of each other.
On a side note, you seem distrust simulations. Simulations represent the world which is as good as your model. GIGO applies. But if a simulation shows the potential for trouble from a simplified, but validated model, you can be sure as hell that the problem will pop up in the real world. The additional complications in a real world transformer compared to the somewhat simplistic model will not make found issues go away.
Currently, the whole electronics industry relies on simulations, because you cannot breadboard a 24 bit audio DAC chip or an octo-core 2 GHz ARM CPU and probe the individual transistors to see why it doesn't work. Transformers also aren't rocket science and can be simulated with great accuracy. If not, we'd see a lot more transformers blow up in our power networks, or high frequency SMPSes go haywire.
What I am trying to explain the our test are able to check the trafo ONLY.
Without any other facts.
Tha ratio 1:20 on FIAT case , but also for Hammond, is the start point and if I put the 6 ohm as load ( secondary acting primary) I must connect the 2500 ohm on primary ( acting secondary) thats' all.
This is pure thory applied on a real trafo.
Any other considerations are out from my intent; then, when I found the real specs I will check the entire circuit, tube inclued that has a own charatheristc that varying from one brand to another, as we know. So we can have different results.
And I have posted them that are compliant with the trafo tests.
All in the real world not simulation
Walter
Without any other facts.
Tha ratio 1:20 on FIAT case , but also for Hammond, is the start point and if I put the 6 ohm as load ( secondary acting primary) I must connect the 2500 ohm on primary ( acting secondary) thats' all.
This is pure thory applied on a real trafo.
Any other considerations are out from my intent; then, when I found the real specs I will check the entire circuit, tube inclued that has a own charatheristc that varying from one brand to another, as we know. So we can have different results.
And I have posted them that are compliant with the trafo tests.
All in the real world not simulation
Walter
No, you measure the transformer's frequency response in a circuit that has different impedances from the target circuit. The frequency response you measure is therefore not representative to (and probably worse than) the performance of the transformer in the target circuit. Why do you want this?
If you only replace this 2500 Ohm load on the primary (which is now acting as the secondary) with a load that is about equal to the driving tube's internal resistance at the chosen operating point (about 700 Ohms for a 300B), the resulting frequency response will become representative to the response achieved by the transformer in the target circuit. It follows from simple theory and is backed by a valid simulation as I've shown you. Why don't you want this?
This sounds like a cop-out because you refuse to accept the above points. They're true, as I've shown you. You have claimed that you can characterize a transformer with the circuit you've shown, I pointed out the error in the method and how it affects the results regarding the frequency response. I also suggested a simple modification (change ONE resistor) to correct this and make the measured response representative to the in-circuit response that you're going to get. Any remaining differences are not worth fussing about.
Yes, tubes are variable. Rp of the 300B may vary between 500 Ohms for a particularly strong tube, to about 1000 Ohms for a tired old specimen. But these extremes are still is a lot closer to the representative value of the source resistance (which is 700 Ohms) than the currently prescribed value of 2500 Ohms.
Can you post a direct comparison on measured results of your transformer using your characterization method versus actual in-circuit measured performance. I mean, side-by-side, in a single post? I actually am too lazy to go wading through different threads in order to find what I'm looking for.
You keep dismissing the simulation. However, I did not pull the simulation parameters out of one of my bodily orifices. The simulation is based on datasheet values of an actual transformer. I have measured the parameters used in the simulation on my transformer and they match with the values in the datasheet. Also, the in-circuit performance of the transformer (driven by an actual 300B) matches with the performance as given in the datasheet. The simulation results match both with the stated performance in the datasheet and with the achieved performance in reality. Is this still not enough validation for you?
I confirm what I wrote. word by word.
In every case I hope someone other can partecipate to this thread.
In addition I can't continue to discuss with someone that show only a virtual test .
I will happy to continue to discuss with you when you are able to show the real test.
In this case, as I id in last weeks, you have to bought a FIAT trafo, an Hammond trafo and a Sowter trafo, then you can arrange your test set following your theory and finally you are able to demonstrate what I have wrong.
The simulation is fine only if it is followed by a practical stuff, it may give you an indication only.
Walter
In every case I hope someone other can partecipate to this thread.
In addition I can't continue to discuss with someone that show only a virtual test .
I will happy to continue to discuss with you when you are able to show the real test.
In this case, as I id in last weeks, you have to bought a FIAT trafo, an Hammond trafo and a Sowter trafo, then you can arrange your test set following your theory and finally you are able to demonstrate what I have wrong.
The simulation is fine only if it is followed by a practical stuff, it may give you an indication only.
Walter
As you want.
I can say that you haven't understand the problem from the beginning.
And this is a can be a valid opinion as yours.
Or not?
I am waiting you tests.
(this remember me another similar discussion where someon spoke with virtual facts against real !!)
I can say that you haven't understand the problem from the beginning.
And this is a can be a valid opinion as yours.
Or not?
I am waiting you tests.
(this remember me another similar discussion where someon spoke with virtual facts against real !!)
Did I miss something?
DCR of primary, and DCR of secondary?
Open circuit (unloaded) voltage measurement to determine actual turns ratio?
Insertion loss?
Are these important too?
I used a vector network analyzer, and special calibrations and special setups
to measure all of that, plus the usual inductance, leakage reactance, resonance, bandwidth and phase under proper driving and proper load resistances, and distributed capacitance measurements. No simulation here, but a lot of math to calculate some of these from the measurements.
As to testing with DC applied, after the other transformer parameters were measured and calculated, then build the amplifier, and measure the -3dB, and -1dB low frequency response, and the Damping factor at mid band. Using that, and the DCR of primary and DCR of the secondary, and insertion loss, you can determine the output tube's operating rp. Then, from the -3dB and -1 dB rolloff, you calculate the inductance under DC.
Whenever possible, calculate from more than just one pair of measured quantities. You can often use another set of measured quantities to verify the first answer.
DCR of primary, and DCR of secondary?
Open circuit (unloaded) voltage measurement to determine actual turns ratio?
Insertion loss?
Are these important too?
I used a vector network analyzer, and special calibrations and special setups
to measure all of that, plus the usual inductance, leakage reactance, resonance, bandwidth and phase under proper driving and proper load resistances, and distributed capacitance measurements. No simulation here, but a lot of math to calculate some of these from the measurements.
As to testing with DC applied, after the other transformer parameters were measured and calculated, then build the amplifier, and measure the -3dB, and -1dB low frequency response, and the Damping factor at mid band. Using that, and the DCR of primary and DCR of the secondary, and insertion loss, you can determine the output tube's operating rp. Then, from the -3dB and -1 dB rolloff, you calculate the inductance under DC.
Whenever possible, calculate from more than just one pair of measured quantities. You can often use another set of measured quantities to verify the first answer.
to 6A3summer
have you seen the results of tests both of trafos and complete circuit?
Please give me a feedback.
Thx
Walter
have you seen the results of tests both of trafos and complete circuit?
Please give me a feedback.
Thx
Walter
waltube,
I am not sure what circuits you are referring to.
Posts 1, 33, and 39?
For the images in post # 1, the impedance curve with the resonance is either calculated,
or the transformer primary was driven from a high impedance (not 6 Ohms), and secondary was unloaded. Or the secondary was driven by a high impedance and the secondary unloaded.
You can see that the resonances are from about 500k Ohms to 1.5 Meg Ohms.
The resonant frequencies look normal for output transformers.
Unloaded, you may see all sorts of things that are not visible when loaded (and not visible with a correctly designed plate impedance driving the primary, and proper load impedance across the secondary, as in a real world amp).
The low frequency roll off of 2 of the transformers indicates lower primary inductance (normal).
This is a single pole rolloff, with the -1 dB point one octave (2x) times higher in frequency than the -3 dB point.
You can see the larger low frequency distortion on 2 of the transformers, since they have less inductive reactances at low frequencies. The tube has to drive the inductive reactance in parallel with the reflected impedance of the load resistor.
We also see the high frequency roll off that is 2 pole, due to the leakage reactance, and the distributed capacitance (normal).
Post # 24
You can see that the basic inductance and inductance at 28 mA is almost identical.
At 71 mA, it apparently has saturated the transformer, and that is the reason the low frequency has been degraded. I would also expect larger intermodulation of low frequencies on the high frequency signals in an Intermodulation test (and so for a music signal).
We also see the high frequency roll off that is 2 pole, due to the leakage reactance, and the distributed capacitance (normal).
My tests:
I no longer have access to the Vector Network Analyzer (it made testing so much easier, even with the special setups I had to make; it did so much of the math for me, and had the graphics too).
That means I had to develop my own test methods, using a signal generator, resistors, digital oscilloscope, and DC Ohmmeter.
After initial transformer tests, I decide which one to use, then build an amplifier.
Then the completed amp is tested using a fast rise square wave generator, Denon Audio Technical CD, CD player, and the digital oscilloscope in scope and FFT modes. The tests include sine waves to see frequency response and harmonic distortion, dual test tones, high/low tones; and 2 high tones for 2nd and 3rd IM test, impulse (frequency response), and more.
As to tube differences, an example of what I do when I can, is use several different manufacturers tubes. I have 4 manufacturer’s 300B types. I compare mid frequency distortion, IM, low frequency response, low frequency distortion, and damping factor. That testing of different tubes tells me a lot about the tube distortion and plate resistance. I may also vary
the tube quiescent current, and or the B+, and re-test.
I hope that helps.
I am not sure what circuits you are referring to.
Posts 1, 33, and 39?
For the images in post # 1, the impedance curve with the resonance is either calculated,
or the transformer primary was driven from a high impedance (not 6 Ohms), and secondary was unloaded. Or the secondary was driven by a high impedance and the secondary unloaded.
You can see that the resonances are from about 500k Ohms to 1.5 Meg Ohms.
The resonant frequencies look normal for output transformers.
Unloaded, you may see all sorts of things that are not visible when loaded (and not visible with a correctly designed plate impedance driving the primary, and proper load impedance across the secondary, as in a real world amp).
The low frequency roll off of 2 of the transformers indicates lower primary inductance (normal).
This is a single pole rolloff, with the -1 dB point one octave (2x) times higher in frequency than the -3 dB point.
You can see the larger low frequency distortion on 2 of the transformers, since they have less inductive reactances at low frequencies. The tube has to drive the inductive reactance in parallel with the reflected impedance of the load resistor.
We also see the high frequency roll off that is 2 pole, due to the leakage reactance, and the distributed capacitance (normal).
Post # 24
You can see that the basic inductance and inductance at 28 mA is almost identical.
At 71 mA, it apparently has saturated the transformer, and that is the reason the low frequency has been degraded. I would also expect larger intermodulation of low frequencies on the high frequency signals in an Intermodulation test (and so for a music signal).
We also see the high frequency roll off that is 2 pole, due to the leakage reactance, and the distributed capacitance (normal).
My tests:
I no longer have access to the Vector Network Analyzer (it made testing so much easier, even with the special setups I had to make; it did so much of the math for me, and had the graphics too).
That means I had to develop my own test methods, using a signal generator, resistors, digital oscilloscope, and DC Ohmmeter.
After initial transformer tests, I decide which one to use, then build an amplifier.
Then the completed amp is tested using a fast rise square wave generator, Denon Audio Technical CD, CD player, and the digital oscilloscope in scope and FFT modes. The tests include sine waves to see frequency response and harmonic distortion, dual test tones, high/low tones; and 2 high tones for 2nd and 3rd IM test, impulse (frequency response), and more.
As to tube differences, an example of what I do when I can, is use several different manufacturers tubes. I have 4 manufacturer’s 300B types. I compare mid frequency distortion, IM, low frequency response, low frequency distortion, and damping factor. That testing of different tubes tells me a lot about the tube distortion and plate resistance. I may also vary
the tube quiescent current, and or the B+, and re-test.
I hope that helps.
I can do that. Yesterday evening I have dug up a 300B and a suitable SE OPT (4k to 4 Ohms), and I will lash up a power stage with them. I will characterize its response on some critical frequencies (-3 dB low frequency, -3 dB high frequency, critical frequency of second order rolloff, corresponding estimated Q) and try to measure the transformer's essential parameters. I will power the stage from a stabilized lab power supply and use fixed bias, so that there are no other components in the setup that might mess up the response.
I will then use your method, and measure the same parameters with the primary (which now acts as secondary) loaded with 4k (as you propose) and with the plate resistance of the 300B (as I propose). I have a curve tracer, so I can determine its actual plate resistance at the chosen operating point. We'll see how the measured properties behave/change under these different conditions, and how they compare to the results from the actual power stage. Just give me a few days, as my agenda is rather packed.
The audio amp that I use for testing may not be entirely op to the task of challenging the HF response of the transformer, but I'll see how far it will get me. I have some other transformers/tubes as well, but since the discussion revolves around a 300B SE stage, this combination looks appropriate to start with.
Last edited:
Thx to 6A3 for the comments. They are well described .
First of all, this thread was open new and some post comes from other thread have been included and maybe there is a little confusion on diagrams.
1- each diagram with the plot of Z and phase with a Montanucci software are with the primary open, this to check exactly the electrical value of each trafo.
2- the diagram of TRAFO ONLY are made with the test set I described ( reverse connection)
3- at this link, post 6
LITZ output trafo
there are diagram related to a tes done with a test set + bias current only related to trafo . In this thread there is a discussion about LITZ OT ( where I am still running !! 🙂 )
In this case the DUT proto from FIAT) was a different one from the last I mentioned here :
why not E88C post 60
the core is bigger and THd at low level is lower.
4- regarding the performace with or without bias, with our test set we decide to check without bias current because the differences are minimal . In addition, in this way, the implementation of the test set is easy and anyone can arrange the circuit quickly and cheaper.
5- the method to check without bias is also related at the fact that if I got a low value of L at low frequency I am sure that with dc the performance are more scarce.
The real world: with 20 H of FIAT at 20 Hz I got 1.3 kohm of Z, with Hammond ( 8.5 H) we have 600 ohms and this is a great difference.
For high frequency the approach is different because the core is less important.
I stop here for the moment
Walter
First of all, this thread was open new and some post comes from other thread have been included and maybe there is a little confusion on diagrams.
1- each diagram with the plot of Z and phase with a Montanucci software are with the primary open, this to check exactly the electrical value of each trafo.
2- the diagram of TRAFO ONLY are made with the test set I described ( reverse connection)
3- at this link, post 6
LITZ output trafo
there are diagram related to a tes done with a test set + bias current only related to trafo . In this thread there is a discussion about LITZ OT ( where I am still running !! 🙂 )
In this case the DUT proto from FIAT) was a different one from the last I mentioned here :
why not E88C post 60
the core is bigger and THd at low level is lower.
4- regarding the performace with or without bias, with our test set we decide to check without bias current because the differences are minimal . In addition, in this way, the implementation of the test set is easy and anyone can arrange the circuit quickly and cheaper.
5- the method to check without bias is also related at the fact that if I got a low value of L at low frequency I am sure that with dc the performance are more scarce.
The real world: with 20 H of FIAT at 20 Hz I got 1.3 kohm of Z, with Hammond ( 8.5 H) we have 600 ohms and this is a great difference.
For high frequency the approach is different because the core is less important.
I stop here for the moment
Walter
Last edited:
to timpert
What are you doing I already sent for trafo only and with the entire circuit.
Look at the link; they match each one
Walter
What are you doing I already sent for trafo only and with the entire circuit.
Look at the link; they match each one
Walter
Only a perfect transformer can be used back-to-front to get the same results. You don't need to measure or simulate a perfect transformer; it simply transforms impedance perfectly. Real transformers are different.
Would it be helpful for waltube to remind us of any interest he has in FIAT or their transformer designs? I seem to recall another thread about Litz-wound transformers, in which apples and oranges were being compared. It would be sad if it turned out that this thread is merely an example of someone with a 'bright' idea using that 'bright' idea to validate the 'bright' idea, while refusing to properly engage with those pointing out the weaknesses of both the idea and the validation.
Would it be helpful for waltube to remind us of any interest he has in FIAT or their transformer designs? I seem to recall another thread about Litz-wound transformers, in which apples and oranges were being compared. It would be sad if it turned out that this thread is merely an example of someone with a 'bright' idea using that 'bright' idea to validate the 'bright' idea, while refusing to properly engage with those pointing out the weaknesses of both the idea and the validation.
to DF96
As I told in a different thread, FIAT give me the proto I ask and I paid for them. That's all.
Every info about the product they have are on FIAT site or can be obtained with an email to them.
The opinion of DF96 are absolutely out of logic and I am waiting the results of tests that he can do if he has the capability.
I can use each real trafo in back to front configuration when it is load with a resitors related with the specs of the stuff and this is the way that is the closest to the real circuits.
If you can't understand this you haven't clear in which way the trafo is workin
Bye
Walter
As I told in a different thread, FIAT give me the proto I ask and I paid for them. That's all.
Every info about the product they have are on FIAT site or can be obtained with an email to them.
The opinion of DF96 are absolutely out of logic and I am waiting the results of tests that he can do if he has the capability.
I can use each real trafo in back to front configuration when it is load with a resitors related with the specs of the stuff and this is the way that is the closest to the real circuits.
If you can't understand this you haven't clear in which way the trafo is workin
Bye
Walter
Last edited:
You wanted me to back up my claims with measurements so I will. You asked for it yourself.