Output transformers, working out output impedance?

ESchap · August 2014

I've got a couple of old WEM 2 x EL84 heads which are fantastic little amps. I'd like to be able to use them together into a 2 x 12 cab loaded with Celestion Blues, one amp into each speaker.

The Woden output transformers in them are marked with a 15 ohm and 3.5 ohm "taps". Problem is the Blues I have are the 8 ohm version. There are other unused "taps" on the output side of the transformers, but these are not marked ... are they likely to be 8 ohm? and could someone tutor me as to how I'd work it out with a meter?

ecc83 · August 2014

Hopefully someone familiar with the amps or transformers will pitch in?

Meanwhile, you say "other taps", plural? I would expect only one more. Normally you have common, 4, 8 and 16. Check with a meter to see if these taps are 1) continuous and 2) connected to the other, known taps.

You MIGHT be able to work it all out by resistance but generally secondaries are so low that it is hard to tell them apart.

Best way is a 15R load and a constant signal and read the volts. We can go into that more deeply if need be.

Push comes to it, run 8 Ohms on 15 tap, no biggie!

Dave.

ICBM · August 2014

Are the other taps continuous with the two labeled ones? If so you should be able to work it out roughly from the DC resistance of the coil, bearing in mind what Dave said about it being very low - you can usually tell them apart though. If one is higher than the 3.75 and lower than the 15 it's an 8 (7.5) and if one is higher than the 15 it's either a 32 (possible on a PA amp) or 100V line (ditto).

ESchap · August 2014

Many thanks chaps. I'll have a look tonight and report back. Interestingly the schematic for the amp only shows a 15 tap but not a 3.5. though this schematic is a couple years younger than the amps, it seems correct in every other way.

As a matter of interest what is the function of the 0.002uF caps and 4.7k resistors around the output tube side of transformer?

ecc83 · August 2014

ICBM said:

Are the other taps continuous with the two labeled ones? If so you should be able to work it out roughly from the DC resistance of the coil, bearing in mind what Dave said about it being very low - you can usually tell them apart though. If one is higher than the 3.75 and lower than the 15 it's an 8 (7.5) and if one is higher than the 15 it's either a 32 (possible on a PA amp) or 100V line (ditto).

Oooo! Did not think about 100volt taps and me having done quite lot of work on 100V line PA systems.

BTW, anyone know why the Yanks use 70.7 volts? Seems a bit pointless and just makes the sums harder!

Dave.

ICBM · August 2014

Has to be something to do with it being 50 times the square root of 2, but what exactly escapes me...

The caps and resistors on the primaries of the OT are Zobel filters, which flatten the impedance curve of the OT and speaker and makes it easier for the valves to produce a wider bandwidth (closer to a resistive load). You often see it in hi-fi, PA and sometimes bass amps, including solid-state ones, but rarely in guitar amps.

ESchap · August 2014

Thanks @ICBM I think these were originally meant to be bass amps but perhaps PA also? Makes sense then. So are the Zobel networks not used in Guitar because they have a bad / limiting effect? or is it they are just perceived as not needed and therefore reduce the component count? Whichever, I really like these amps for guitar, really harmonically rich clean and vintage small Marshall when driven.

ecc83 · August 2014

I saw those CR networks (called "snubbers" when used in SMPSUs and line timebases) across the really cheap transformers in telly sound outputs.

IRRC they were there to cut 3rd harmonic distortion which might actually be the case since the speakers were almost always also small and cheap and therefore tended to have a good but spiky HF response?

Dave.

ecc83 · August 2014

ecc83 said:

I saw those CR networks (called "snubbers" when used in SMPSUs and line timebases) across the really cheap transformers in telly sound outputs.
IRRC they were there to cut 3rd harmonic distortion which might actually be the case since the speakers were almost always also small and cheap and therefore tended to have a good but spiky HF response?
And! Forgot. Once in a very long while the cap would short and the crap carbon comp' resistor cook and go low and give almost no sound output.

Dave.

ICBM · August 2014

I presume they're not generally used in guitar amps because the do smooth the HF response, as well as costing money! Quite a lot of vintage amps that are derived from hi-fi(ish) circuits have them though. The Fender Pro Junior also does, fairly uniquely for a modern valve guitar amp as far as I know. (Single cap and resistor across the whole primary, rather than a pair to the centre tap.) It might explain its nicer, more open and much less boxy/middy tone than the bigger Blues Junior which doesn't.

ecc83 · August 2014

ICBM said:

I presume they're not generally used in guitar amps because the do smooth the HF response, as well as costing money! Quite a lot of vintage amps that are derived from hi-fi(ish) circuits have them though. The Fender Pro Junior also does, fairly uniquely for a modern valve guitar amp as far as I know. (Single cap and resistor across the whole primary, rather than a pair to the centre tap.) It might explain its nicer, more open and much less boxy/middy tone than the bigger Blues Junior which doesn't.

Yeah, meant to say "guitar amp peeps" probably like the distortion and the speaker would filter out the worst of it.

As for cost, tellies were as cheaply made as possibly (who remembers Radio Rentals?) and a cap and an R was cheaper than a decent OP traff!

Dave.

jpfamps · August 2014

Generally the Zoebel network (sometimes referred to as a conjunctive filter), is used to try to present a near constant impedance to the power valves as a way of compensating for the rising impedance of a real world load speaker rather than a resistive load.

This technique was popular before negative feedback was habitually employed to linearize the output.

It will have the effect of shunting high frequencies to AC ground, so will reduce HF response.

Dr Z has used this in some of his amps.

One of the problems with an overdriven guitar amp is that copious high frequencies are produced when compared with music reproduction, thus a resistor with significant power handling is required, and I guess it's seen as an unnecessary expense for a guitar amp.

The Zobel network in a solid state amps performs a different function: it is required to ensure stability into all loads, although surprisingly the exact mechanism by which this works seems obscure.

ecc83 · August 2014

jpfamps said:

Generally the Zoebel network (sometimes referred to as a conjunctive filter), is used to try to present a near constant impedance to the power valves as a way of compensating for the rising impedance of a real world load speaker rather than a resistive load.

This technique was popular before negative feedback was habitually employed to linearize the output.

It will have the effect of shunting high frequencies to AC ground, so will reduce HF response.

Dr Z has used this in some of his amps.

One of the problems with an overdriven guitar amp is that copious high frequencies are produced when compared with music reproduction, thus a resistor with significant power handling is required, and I guess it's seen as an unnecessary expense for a guitar amp.

The Zobel network in a solid state amps performs a different function: it is required to ensure stability into all loads, although surprisingly the exact mechanism by which this works seems obscure.

I no longer have the book but I recall a Mullard transistor audio circuits collection gave the operation of the Zobel network as cancelling out (to a degree) the inductive component of the speaker which could (they said) lead to transistor breakdown.

I shall give Amazon a go now I am reminded of the book!

Dave.

ICBM · August 2014

Yes, that was what I understood the purpose to be, to present a constant impedance to the output devices rather than the inductive speaker and/or transformer.

I also suspect they're sometimes used as a band-aid to fix flyback voltage problems that the designers can't cure any other way. My reasoning is that I have come across several (more than one, although I don't remember exactly how many) where the *cap* has failed - shorted - sometimes also burning out the resistor but sometimes not - it just nearly mutes the amp if the resistor survives. It must be that way round since otherwise the resistor failing due to excess HF power would protect the cap. For this to happen the cap has to be exposed to excess voltage, and it's often rated at 1KV or more. One of these was a Pro Junior, and it baffled me how it could have happened at the time since there should be no way over 1KV can be present in an amp with a B+ of only 320V.

ESchap · August 2014

Hmmm ... just opened both up and they're different ... not been inside one of them for a couple of years. The earlier one with the Woden Transformer seems to be outputting from the 3 ohm tap? might account for it sounding a bit different to the other as I've been running it into 16 ohms! The other is a 1964 and possibly has a replacement transformer, but it's quite a bit bigger than the Woden, it appears to be running at 16ohms, the vacant tap reads at half the value of the 15 ohm tap on my meter so I'm assuming it's 8.

The Woden transformer's Zobel network is as per the schematic, with what look like 2 or 3 watt wire wound 4.7k resistors and 0.002uF caps of a type don't know.

The other one has the network as well but it has 33k 1watt CC resistors and 0.022uF 400v Mustard caps!

Considering the differences in values, they do sound remarkably similar with the Woden transformer sounding a little quieter and perhaps a tad more polite when cranked, though if it is set at 4ohms that might be accounted for by being put through a 16ohm speaker?

The '63 is all Hunts caps with a couple of large Wima's and I've just noticed that all the black ones are all cracked :-S .. though the amp operates OK. The '64 is all Wima and mustard caps, with a couple of large red Hunts.

Feel I've taken a few steps backwards now. Suspect a visit to ICBM's headquarters is required.

1963

The black sheathed red and black wires go to the output jack

1964

The black gunge is most likely the result of the mains cap blowing sometime in the past.

.

ecc83 · August 2014

Hunts caps were famous for going leaky/short in tellies (espc' IF amp decouplers)

Wimas were famous for going short in Grundig tape recorders.

Mustards are famous for never failing at all!

Dave.

ICBM · August 2014

Something has certainly been on fire in the second one! If it wasn't for the soot on the OT I would assume it had been the previous one… but it must have been after it was changed. Seems like this amp has a bit of a history of it! It's a reasonable assumption that the Zobel network components aren't original since they're far too clean, so it's possible the values aren't right if they were copied from damaged or unreadable originals.

4K7/.002uF are more typical values than 33K/.022, but they might actually work out similar in practice.

As Dave says the black Hunts caps are terrible for failing, they've almost all cracked now. Wimas aren't great either although they seem to be sought-after for some reason! Mustards almost never fail - as far as I can remember the only ones I've seen which have are in the PI to power valve position in fixed-bias amps where they will usually have over 500V on them (although are 600V caps so *should* be able to take it) - but only a handful.

jpfamps · August 2014

ecc83 said:

jpfamps said:

Generally the Zoebel network (sometimes referred to as a conjunctive filter), is used to try to present a near constant impedance to the power valves as a way of compensating for the rising impedance of a real world load speaker rather than a resistive load.

This technique was popular before negative feedback was habitually employed to linearize the output.

It will have the effect of shunting high frequencies to AC ground, so will reduce HF response.

Dr Z has used this in some of his amps.

One of the problems with an overdriven guitar amp is that copious high frequencies are produced when compared with music reproduction, thus a resistor with significant power handling is required, and I guess it's seen as an unnecessary expense for a guitar amp.

The Zobel network in a solid state amps performs a different function: it is required to ensure stability into all loads, although surprisingly the exact mechanism by which this works seems obscure.

I no longer have the book but I recall a Mullard transistor audio circuits collection gave the operation of the Zobel network as cancelling out (to a degree) the inductive component of the speaker which could (they said) lead to transistor breakdown.
I shall give Amazon a go now I am reminded of the book!

Dave.

Indeed, however Doug Self's investigations show evidence of VHF oscillation when the Zobel filter is removed by an unspecified mechanism, ie not Nyquist instability.

ecc83 · August 2014

jpfamps said:

ecc83 said:

jpfamps said:

Generally the Zoebel network (sometimes referred to as a conjunctive filter), is used to try to present a near constant impedance to the power valves as a way of compensating for the rising impedance of a real world load speaker rather than a resistive load.

This technique was popular before negative feedback was habitually employed to linearize the output.

It will have the effect of shunting high frequencies to AC ground, so will reduce HF response.

Dr Z has used this in some of his amps.

One of the problems with an overdriven guitar amp is that copious high frequencies are produced when compared with music reproduction, thus a resistor with significant power handling is required, and I guess it's seen as an unnecessary expense for a guitar amp.

The Zobel network in a solid state amps performs a different function: it is required to ensure stability into all loads, although surprisingly the exact mechanism by which this works seems obscure.

I no longer have the book but I recall a Mullard transistor audio circuits collection gave the operation of the Zobel network as cancelling out (to a degree) the inductive component of the speaker which could (they said) lead to transistor breakdown.
I shall give Amazon a go now I am reminded of the book!

Dave.

Indeed, however Doug Self's investigations show evidence of VHF oscillation when the Zobel filter is removed by an unspecified mechanism, ie not Nyquist instability.

I would like to follow Doug's findings up, one of the sensible audio guys! Do you have a link?

Re Zobel networks. A lot of cheap, IC based power amps use a really tiny resistor and under overdrive or perhaps a bit of HF instability it burns out, as it is often slammed up against the cap, it cooks that which shorts and the rest of the amp is history!

Dave.

jpfamps · August 2014

ecc83 said:

jpfamps said:

ecc83 said:

jpfamps said:

Generally the Zoebel network (sometimes referred to as a conjunctive filter), is used to try to present a near constant impedance to the power valves as a way of compensating for the rising impedance of a real world load speaker rather than a resistive load.

This technique was popular before negative feedback was habitually employed to linearize the output.

It will have the effect of shunting high frequencies to AC ground, so will reduce HF response.

Dr Z has used this in some of his amps.

One of the problems with an overdriven guitar amp is that copious high frequencies are produced when compared with music reproduction, thus a resistor with significant power handling is required, and I guess it's seen as an unnecessary expense for a guitar amp.

The Zobel network in a solid state amps performs a different function: it is required to ensure stability into all loads, although surprisingly the exact mechanism by which this works seems obscure.

I no longer have the book but I recall a Mullard transistor audio circuits collection gave the operation of the Zobel network as cancelling out (to a degree) the inductive component of the speaker which could (they said) lead to transistor breakdown.
I shall give Amazon a go now I am reminded of the book!

Dave.

Indeed, however Doug Self's investigations show evidence of VHF oscillation when the Zobel filter is removed by an unspecified mechanism, ie not Nyquist instability.

I would like to follow Doug's findings up, one of the sensible audio guys! Do you have a link?
Re Zobel networks. A lot of cheap, IC based power amps use a really tiny resistor and under overdrive or perhaps a bit of HF instability it burns out, as it is often slammed up against the cap, it cooks that which shorts and the rest of the amp is history!
Dave.

This info is in Doug's book "Audio Power Amp Design" 5th Edition in the discussion of compensation and stability.

I don't have the 6th edition of this book (yet), but I'm sure that this info will be there too and possibly even expanded on.

ESchap · August 2014

Well, Zobel networks aside, I'd like to thank @ICBM who had a look at these at Lunchtime for me. The Woden transformer is 4ohm only and the other is 16 and 8 ohms. Not exactly what I wanted, but 4 ohms into 8 should be OK, certainly better than the 16 ohm cab I've been running the amp into for the past 2 years! Found a cathode electrolytic cap that looks ready to pop and a fair bit of remedial work to do changing all those peky Hunts caps, better warm up the old iron ...

Howdy, Stranger!

Categories

In this Discussion

Become a Subscriber!

Output transformers, working out output impedance?

Comments