• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Basic tests and measurements for OPTs

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I have two "project" amps right now - both PP EL34 + 12AX7 driver type amps. One is a Silvertone, and the other came out of a RCA/Westinghouse stereo tuner. Besides figuring out impedance ratios and inductance, I want to do some basic measurements on the output transformers of each. Just to get an idea of what quality of transformers I am dealing with. They are P-P btw.

I guess frequency response and distortion tests would be in order ? Now here's the thing - I don't have any fancy measurement equipment - not even a wide bandwidth oscilloscope or spectrum analyzer right now, except for a PC based measurement setup i.e. not much bandwidth or power output. Well for power output I guess I could use another amp but measurements would be limited to less than 50khz. Will be using popular PC programs like RMAA or VA.

Anyway, so how do I go about "testing" these transformers with what I have ? Here's what I was thinking - I could do a loopback from the souncard's Line-out to Line-in, with a power amp and the transformer within the loop. So it would look like -

Soundcard LineOut->PowerAmp->Transformer Pri : Transformer Sec->Load->Soundcard LineIn.

The soundcard's line-out would go to the power amp and the transformer's primary will be connected to the amp's speaker output. (I could do tests for the full primary and also for each half of the primary). A load will be put on the secondary and that will be connected to the soundcards line-in.

Then run typical loopback back tests for frequency response and THD for different output/power levels. What'd you think ? Anybody see any problem with this ?

TiA!
 
I do not know much about how to evaluate OPT's but i think the PC soundcard response will normally be limited to 18-20 KHz and therefore severely limiting the usefulness in this case.

I have not tried if an external sounbox with 96KHz sample rate might be better. I usually borroow a tone generrator and a scope when I want to look at tube amp 10K squares.

SveinB.
 
Transformer testing

Your scheme is good and in fact very similar to what I use for testing. One thing, though - the frequency response and distortion is quite dependent on the source impedance. If I am testing an 8-ohm transformer, I put an 8-ohm dummy-load resistor in series between the power amp and transformer secondary. I then put a load resistor across the primary that approximates the source impedance in the amplifier. Usually something like one-half the tube Rp is good. Both the 8-ohm resistor and the primary load resistor have to handle the maximum power you plan to use. Also, the power amplifier has to be able to supply double the power you want, since half of it gets soaked up in the 8-ohm resistor. I use an old Crown DC-300A amplifier for my tests - it is rugged, has response down to DC, and isn't used in my listening system;)

To test high frequencies, a quick test is to put a square wave in. However, unless you use a really high performance sound card that can sample at something like 192KHz, you will be limited by what you can see at high frequencies.

At low frequencies, you will see transformer saturation show up as notches in the sine wave. A good test is to pick your desired power level, then lower the frequency until the saturation starts to happen. On a good transformer, this should be 40Hz or less.

The distortion caused by the transformer at low signal levels is subtle, but IMHO, contributes the most to the "sound" of the transformer. This distortion actually increases as the signal level goes down. A good way to see this is to look for odd-order distortion products while driving the transformer at a low frequency that is not harmonically related to the power line hum. I often use 28 Hz. You will need a good 24-bit sound card and an FFT program to see the distortion. Even-order distortion products are meaningless here, since they depend on the past magnetization history of the transformer, and the distortion mechanism generates odd-order products anyway.

I hope this helps,

- John Atwood
 
Re: Transformer testing

John, thanks for your reply. I will keep your comments in mind when I do the tests.

JohnAtwood said:
One thing, though - the frequency response and distortion is quite dependent on the source impedance. If I am testing an 8-ohm transformer, I put an 8-ohm dummy-load resistor in series between the power amp and transformer secondary. I then put a load resistor across the primary that approximates the source impedance in the amplifier.

so hookup the transformer in 'reverse' ? i.e. secondary connected to the power amp output ? Would that have any implications considering the flow of power is going in the other direction ? It shouldn't but...

I guess then the large voltage ratio would result into a big voltage swing at the primary, so I'd need a voltage divider to ensure I dont overload(smoke) the soundcard's input ?
 
trans test

I forgot to mention that a frequency-compensated voltage divider is absolutely essential when measuring the output. I know this the hard way - I've twice blown the inputs on my Audio Precision System Two when working with tube circuits:( . Since a sound card has little or no input attenuation, you will have to make a voltage divider that brings the worst-case input down to 1Vrms or so. There should be a small adjustable capacitor across the input resistor that can be adjusted to give flat frequency response (it is adjusted just like the compensation on a 10:1 scope probe).

By the way, you can hook the transformer in the forward direction (amp output to primary, soundcard input to secondary), but you will still need the resistors: several thousand ohms in series with the primary and 8 ohms (or whatever the tap impedance is) across the secondary. The problem here is that you will have a hard time driving the transformer into saturation. Running it "backwards" makes it much easier to do this. The transformer doesn't care which way the power is flowing.

- John
 
Based on John's advices I will try the following:

Connect the power amplifier with 8 ohms series dummy load resistor to the output of my PP tube amplifier, and measure the voltage at the anodes, with the tube amplifier switched on, but no signal at its input. This way the OPT will see real load (the internal resistance of the triodes) on the primary. The tube amplifier has no negative feedback, so hopefully this will not influence anything. I just have to construct a 10:1 compensated divider...
 
Well, I tried out the above idea, i. e. feeding power into the secondary of the OPT of a PP tube amp through 8 ohm. It worked and I could measure the voltage on the primary and on the secondary, and I could calculate the turns ratio of the OPT from the values. However, there is an interesting phenomenon:

The voltage at the speaker terminals of the tube amplifier is a given value (11.25 V) if the amplifier is turned off. This voltage drops to 8.00 V when I turn on the filament voltage (no plate voltage applied yet). If the plate voltage is turned on, the voltage on the speaker terminals drops again to the half of the previous value, 4.00 V. This is normal I believe, because the output resistance of the tube amplifier is 8 ohms and half voltage drops on the external series 8 ohms dummy load. The amplifier runs in open loop, without NFB.

What I can not explain, why is the voltage on the speaker terminals different with cold amplifier and with the filament voltage applied only? Could it be the internal filament-cathode diode that is turned on and may have some shunting effect?
 
No, that was not the internal diode. I raised the filament potential well above the cathode voltage, and the effect is still present. But this mysterious load effect decresed when I disconencted the g2 (which is tied to the anode), still no plate voltage applied. And it decreased further when I disconnected the anode, leaving the OPT open. Here is the question: why is the triode loading the transformer when it has no plate voltage, only filament voltage?
 
Administrator
Joined 2004
Paid Member
Dave Cigna said:


Even if B+ is not applied, the signal coming in backwards through the OT is driving the each plate positive during 1/2 cycle, thus loading down the other amp.

Driving the output transformer in reverse in a pp power amplifier with filament power, but no B+ or bias applied results in a full wave center tapped rectifier circuit instead of your normal output stage. (With the results noted by Dave above.) Note that carried to an extreme this may damage electrolytics in your power supply if the voltage is an appreciable % of the cap's rating due to them being reverse biased.
 
Hm, that makes sense. So this is not a correct measurement. Actually I wanted to measure the internal resistance of the output tubes in the circuit and determine the optimal turns ratio, where the amplifier transfers the highest power onto the load. What about the following method?

Leave the secondary of the output transformer open. Drive the tube amplifier so that is produces some reasonable voltage at the secondary of the output transformer. Then attach a dummy load at the secondary. The internal resistance can be calculated from the readings and from the turns ratio (squared). I know that running the tube amplifier unloaded is dangerous for the OPT primary, so I chose an output voltage of less than 10 V. The amplifier operates in open loop.
 
Normally you don't want to load a transformer durring measurements.... You measure it unloaded....
The main numbers you need are primary inductance... This changes according to flux density, so you graph the inductance vs. power ....
The other key point is the transformers resonance. You get this by driving the secondaries with a power amp and sweep the frequency till you get a peak.... Once you get the peak frequency, you can then determine the capacitance and leakage inductance...
Once you know the leakage inductance, that number stays fixed on a transformer since that is a function of winding geometry...
The problem next is remeasuring the the capacitance at different drive levels, since the capacitance will vary according to drive level, similair how the inductance varies.... This is due to the changing AC voltage gradient on the dielectric material in the transformer...The dielectric constant is actually not constant and has a slope that changes to applied AC voltage gradient...
You will see tha the transformer's resonance peak will drop in frequency as the power level is increased...
Once you have these numbers you can then mathamatically apply the plate resistance of the tubes used to determine the frequency response of the transformer at various power levels...

Chris
 
Administrator
Joined 2004
Paid Member
oshifis said:
Hm, that makes sense. So this is not a correct measurement. Actually I wanted to measure the internal resistance of the output tubes in the circuit and determine the optimal turns ratio, where the amplifier transfers the highest power onto the load. What about the following method?

Leave the secondary of the output transformer open. Drive the tube amplifier so that is produces some reasonable voltage at the secondary of the output transformer. Then attach a dummy load at the secondary. The internal resistance can be calculated from the readings and from the turns ratio (squared). I know that running the tube amplifier unloaded is dangerous for the OPT primary, so I chose an output voltage of less than 10 V. The amplifier operates in open loop.

To determine the source impedance of the amplifier this is a technique that will work, I would keep the voltages very moderate. (Just a couple of volts with an accurate volt meter should be more than sufficient.) Note that there is an excellent note on this measurement technique on the Audio Precision website. Generally used to determine source impedance in SS feedback amplifiers, it will work just as well with a tube amplifier whether or not it uses global feedback. Of course you can calculate it as cerrem has indicated as well, but doing the measurement will be more accurate and is actually pretty easy.

The equation Rout = (Vout - Vload) x Rload/Vload where Rout is the amplifier source impedance at the speaker terminals, Vout is the unloaded output voltage, Vload is the loaded output voltage, Rload is your resistor load.. Note that this includes the dc resistance of the secondary which you should subtract from the measurement before multiplying by the impedance transformation ratio (turns^2) to determine the approximate source impedance of the amplifier. Frankly the secondary number is usually the one of most interest. I've not tried this out in practice so I can't say how well it will actually work, I may have missed something obvious along the way.

In normal applications this would be done at 10Vrms, but given the high source impedance here much lower voltages can be used with good accuracy.
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.