OT Inductance vs High Freq Response

[tbryanh]

I picked up an old vacuum tube amp that uses a 600v plate supply in the PA.

I heard that the OTs for such a high voltage can have more inductance (more turns) than for a lower voltage which can cause a loss of high frequency response.

I also heard there are special transformers to be used in situations like this to avoid losing the highs.

The infomation was on some website where the person was speaking in general terms without giving any specific examples or information.

I am wondering if I can find out info about the OT in my amp. I suppose I could remove the transformer, put a dummy load on the secondary, and measure the inductance of the primary?

I suppose I need to use the plate resistance along with the transformer inductance and other circuit component info to draw up an equivalent circuit of the output stage?

Then I can analyse what kind of filtering effect the final has on the freq response.

Seems a little complex but not too far out of reach, but of course I have never done this before, so I am hoping for a little help.

One other thing I was thinking. Maybe analyzing just the output stage is not enough. If the highs did drop there, I suppose they could be boosted in an earlier stage to make up for this. If this is the case, I am wondering if the sound would be good in a situation like this?

I plan to convert the amp into a guitar amp. It is an old Stromberg Carlson amp I think. Actually it is missing all the covers and stuff, so I am not sure what it is.

Thanks

[R.G.]

There are about three layers needed to answer that seemingly simple question.

1. Transformers

I heard that the OTs for such a high voltage can have more inductance (more turns) than for a lower voltage which can cause a loss of high frequency response.

This is a mishash of what transformers actually do. Go find and read a reference on transformer modelling. In short, a transformer is modelled as an ideal, lossless, perfect frequency response device doing the transforming, and external imperfections modelling the realness of the actual device.

The imperfections are the resistances of the primary and secondary wire, the non-infinite, lossy, and saturating inductance of the primary, the nonlinear resistive losses of the core eddy currents, the leakage inductance from primary to secondary, and the distributed capacitances in the winding(s).

Frequency response is limited on the low end by the limited inductance of the primary. In turn, this is limited by the core saturation and the input frequency. A real core can only be magnetized to a certain amount of volt-second product without saturating or having otherwise unacceptable losses. The lowest frequency of interest must be coped with, and the highest voltage applied must be coped with. That sets the volt-seconds to be withstood, and that chooses the core cross sectional area.

Presumably the fidelity is of interest, so you must choose a primary inductance which "eats" less than half (and often much less than half) the power provided to the transformer at the lowest frequency and impedance level, so the inductance is set by those numbers. With the minimum inductance and minimum volt-seconds set, the primary number of turns is a straightforward calculation. That is the fewest number of turns which can do the job. Inductance is proportional to the square of the number of turns.

The import of that last is that high voltage pushes the necessary inductance up, which pushes the number of turns up. The secondary turns have to go up as well to preserve the transformer ratio. The LEAKAGE between primary and secondary matters because flux which does not couple from primary to secondary looks like an imperfection-inductor in series with the primary. And that is the connection between voltage and frequency response. Higher voltages require more inductance, requires more turns, and you get more leakage as a side effect.

I also heard there are special transformers to be used in situations like this to avoid losing the highs.

The special transformers have the same size core and same number of turns in primary and secondary, but have the primary and secondary interleaved: a portion of the primary is wound, then part of the secondary, then part of the primary, then... until all of both primary and secondary are wound. The interleaving reduces the leakage. It also INCREASES the capacitance losses, which cause high frequency losses, too. There are no simple answers in transformer design.

2. Amp class effects on transformers and vice versa
The stuff above is for a constant-AC situation. In guitar amps, we use class AB a lot, and that means that one of the two tubes on the output transformer turns off on big excursions in each direction. When part of the primary is turned off, it can't contribute to power transfer, so now you have two primaries that both have to be interleaved independently into the whole secondary, and for reliability reasons also have to be interleaved with each other. More capacitance, more treble loss again. Harder transformer design and winding.

3. The rest of the circuit.
Which is all you get to play with unless you can wind your own.

I am wondering if I can find out info about the OT in my amp. I suppose I could remove the transformer, put a dummy load on the secondary, and measure the inductance of the primary?

You can do the classical transformer tests - open circuit primary test gives you primary inductance. Shorting the secondary and measuring inductance gives you the leakage from the winding you are using as a primary to the shorted winding. In general the leakage from each half primary to the secondary is not the same, unless strenuous measures are taken in winding to make them the same.

I suppose I need to use the plate resistance along with the transformer inductance and other circuit component info to draw up an equivalent circuit of the output stage?
Then I can analyse what kind of filtering effect the final has on the freq response.

Yes. Note that the nonlinearities in the primary inductance with the size of the signal put on it and the amount and balance of the DC currents through it will make this a very complicated task.

One other thing I was thinking. Maybe analyzing just the output stage is not enough. If the highs did drop there, I suppose they could be boosted in an earlier stage to make up for this. If this is the case, I am wondering if the sound would be good in a situation like this?

Yes, all of the above. The actual analysis is very complicated. It can be pre-boosted if and only if the amp has enough gain and power available to force the transformer iron to do something it does not naturally want to do, and if that amount of "forcing" does not damage the transformer.

Example: your transformer has a drop of 12db at 10kHz and you want to pre-correct that. That drop actually means that the iron and copper is only transmitting 1/4 of the power you're trying to feed it. If you feed it four times the power at that frequency, it will get the 1x power you wanted through to the secondary. The question is - where did the 3x power that was lost go? Is is heating the transformer? The output tubes? Do you have a 3X power reserve in your power supply to supply that?

My advice? Don't over-analyze this. Do the amp conversion, measure what you get, and only then if you don't like it go out and jump off the analysis cliff. It's deep water that you haven't been swimming in before. You may want to, but make sure you want to before you get in too deep.

[tbryanh]

Thanks for the great response.

"Go find and read a reference on transformer modelling"
I've been trying. I've spend money on eBay for vintage books, and I've bought books from Lindsay's Technical Books http://www.lindsaybks.com. I would love to have a great book on transformers. Please refer me.

"volt-seconds"
I'm not sure what this term means. Please explain.

"...high voltage pushes the necessary inductance up, which pushes the number of turns up. The secondary turns have to go up as well to preserve the transformer ratio."
This statement appears correct and has helped me gather some of my thoughts. Here is what I have to support it.

It appears that the primary impedance is increased to increase the ac voltage gain. If the ac plate current stays the same and the plate supply voltage increases, the ac voltage gain of the stage does not change. But if the ac impedance of the primary increases at the same time, then the ac gain of the stage increases.

The plate supply voltage is increased to increase output power. More output power means a higher ac voltage on the secondary. If the number of turns on the secondary stayed the same when the plate supply votage increased and the number of turns on the primary increased, the ac output voltage would stay the same. So the number of turns on the secondary needs to increase too.

[tbryanh]

Now I have decided that I might like to lower the plate voltage and use different output tubes. The existing tubes are 8417. I have heard these are a nice sounding tubes, but they were only made for a short time just before transistors became mainstream and can be expensive and hard to find.

I am thinking that other output tubes will need a lower plate supply voltage, probably about 300v. I don't know if this is true or not, but lets assume it is.

So now I am wondering if I can use the existing output transformer at a lower plate supply voltage. As long as I don't exceed the current rating of the windings, it seems it should be OK.

To prevent clipping, if the plate supply voltage decreases, the ac plate current must decrease too. (Bias will probably need to be adjusted too.) This means there is less power in the primary and secondary. OK, I can live with a lower output power.

Now I am wondering about the output impedance. Of course the new tubes will have a different ac resistance, but for the sake of this discussion, lets assume they don't. In fact, to keep things real simple, lets assume I use the same tubes.

It seems that the output impedance has not changed.

This approach appears to be power scaling. I have read a little about this in Kevin O'Connor's books, but I do not understand the subject well. This discussion seems to shed some light on it.

It appears the method to reduce output power is to reduce the plate voltage, adjust the bias, and reduce the drive to the grid.

[Amp Kat]

Man tbryanh got a headfull there RG and that was some nice stuff. Transformers are a complex thing and under estimated in complexity and to be honest they are quite boring which is why we all probably envy your knowledge so much. Tbryanh, there are some downloadable links at Technical Books Online.htm. If you scroll down you will see a Transformer book by Reuben Lee called Electronic Transformers and Circuits. You can download that whole book but I'm warning you it has complex formulars and engineering symbols used. That's probably why so many technicians can't relate to some transformer talk as the math,calculations,formulars and algebric phase relationships like vectors come into place. I mean theres four pages of symbols and a dictionary to learn before you even get thru the first chapter so it is quite challenging but do-able IMO. Menno Van DerVeen has a book called Transformers and Tubes and in one of the chapters in a whole subject on frequencies in the complex domain which may explain some of your questions but once again it is very technical stuff. Power scaling: well I've built several and even use the circuit as a fixed voltage regulator for dropping B+ to a desired voltage. It is really a power attenuator that controls voltage as a whole to do it. It also has frequency effects as do all attenuators do and the varables effected which or directly related to frequency in the form of impedance use XC,XL and resistance and once again can get into some pretty complex harry stuff.

[Arthur B.]

Is the present transformer taking off too much of your highs? The transformer in there should be designed to handle at least the frequency range of the human voice.

[Enzo]

My question as well. Two things: First, what is the existing freq erspons of the thing, and how does it fall short? And second, is this for a guitar amp? In which case, how much high end are you concerned about? There is a reason there are no tweeters on guitar amps.

Hifi amps have a wide freq range becuase they are made to faithfully REproduce sound. Guitar amps are a part of the sound production, they are not made to be flat or have a broad freq range. Plug the guitar into a PA system and play through PA speakers to hear what it sounds like full range and flat response - pretty thin and tinny.

[tbryanh]

Thanks for the great responses.

As Enzo indicates, the response of the PA amp is probably wider than that needed for a guitar amp, so high freq response is probably not an issue when the PA is converted to a guitar amp.

I don't know the existing freq response. The 60s stuff is from the golden era, so it is probably good.

I will check the books that Amp Kat mentioned. I've been warned, so I know I might notl get too far with them, but I appreciate the referral, that was important.

Thanks again. This is a wonderful website with a lot of good members.

[Rob Mercure]

So what model PA amp did this come out of? You mention 8417s - which are sweet tubes - and I've got two Bogen M-60A that use these although the Ep shouldn't be 600V. Anyhoo, if you provided an amp model number this would allow anyone familiar with the amp/circuit to provide help.

Rob

[tbryanh]

This thing is just a chassis, sorry.

I saw this company on the internet that claimed they bought all the NOS 8417s when they stopped being made.

They made some product at the time that used them.

They say they eventually ran out of them and then converted their product to some other tube. they did not say what they did to convert it.

Are you worried about finding tubes for your Bogen?
Do you plan to convert to some other tube?

[Enzo]

If your amp chassis has no tubes in it, then your B+ is not 600v. As soon as all the tubes are in there, that voltage will drop considerably. Think if what the engine speed on your car would do if you knocked it out of gear at highway speed. Without the tube load, the voltage will run very high. 8417s tend to last a long time, a seriously heavy duty tube.

[bob p]

i was reading Reich's Principles of Electron Tubes last night, and it answers some of your questions about OT inductance, leakage, etc. its available in PDF format from Pete Millett's site.

[Rob Mercure]

Like Enzo said, the 8417 is a rugged tube and my NASA surplus amps still have the original factory tubes. As I've got two of these plus 2 NOS 8417s in the box I suspect that I've got a "lifetime" supply - I'm 53, these amps were made in the late 1960s so they're approximately 40 years old. If I get another 40 years out of the 8417s I suspect that they will sufice for my 93rd birthday party jam <grin>.

And, if I wanted to, what would I convert to? The 6550, et al, is a poor substitute - more grid drive and bias needed and "cold" sounding. The only thing that sounds close are 7591s - and these won't dissipate enough power, or "real" NOS 6CA7s which are becoming as scarce as 8417s (and also also won't dissipate the same power but more than the 7591). Nah, I guess I'll just have to take my chances!

Rob

[Amp Kat]

- I'm 53, these amps were made in the late 1960s so they're approximately 40 years old. If I get another 40 years out of the 8417s I suspect that they will sufice for my 93rd birthday party jam <grin>.

Nah, I guess I'll just have to take my chances!

Rob

Gee Rob ,I hope I'm hanging on with you bud as you got me by 6 but hey, I'll take second chair if you'll save me a spot.

[Rob Mercure]

Hey, you can wheel yourself up there and take the spotlight - I like to share. Anyhoo, since I've played solo longer than I've been in bands - although I find myself in one again after all these years - I like to play a "complicated rhythm" part. Mostly chords with whatever bass or lead licks I can squeeze in to kinda create a "center pivot" that other musicians can bounce off of while I try and keep the whole mess cohesive. So I'll need a lead player and you get bonus points if you can yodel <grin>.

Rob