Edcor 150:600 output tx

[FunkyB]

Id like to use either a Edcor XSM https://www.edcorusa.com/xsmseries or PCW https://www.edcorusa.com/pcwseries as a 1:2 step up output transformer.

Since Edcor doesn't advertise any as being 150:600, and since I'd like to drive the primary unbalanced/single ended with ground as reference, I'm wondering whether I'd be better off using a 600/150 backwards, or if I'd be better off using a 600/600 while only driving the input(+) and the (grounded) C.T.?
And if the second option is the case, should I then leave the input(-) unconnected, or would I be better off grounding it together with the C.T.?

Basically, I'm searching for a 150/600 trafo. Either a 600/150 backwards or using only one half of the primary of a 600/600. Which do you think would be the best solution?

 

[Dylan W]

The best 150/600 tx they make is probably their quadfilar API copy.

https://www.edcorusa.com/xs4400

 

[ruffrecords]

The best 150/600 tx they make is probably their quadfilar API copy.

https://www.edcorusa.com/xs4400

The only problem with this transformer is it is specified for a maximum voltage of 7.5VAC or just under +20dBu.

Cheers

Ian

 

[flaheu]

I have several times used WSM 600/150 backward as output TX and they sounded very good !
Better than their XSM serie in my opinion

 

[joaquins]

The best 150/600 tx they make is probably their quadfilar API copy.

https://www.edcorusa.com/xs4400

The only problem with this transformer is it is specified for a maximum voltage of 7.5VAC or just under +20dBu.

Cheers

Ian

But that's per winding, so with three in series as output it's about +29dBu, so pretty close to what you are probably looking with this transformer.

JS

 

[FunkyB]

I have several times used WSM 600/150 backward as output TX and they sounded very good !
Better than their XSM serie in my opinion

That's good to hear! How did you terminate the secondary. Only resistive, or with a zobel?


In order to get enough output swing, a third option would be to drive the primary balanced (see image). The downfall would be the extra components. A plus could be that the output transistors would only have to deliver half the current.

Given the fact that the output transistors can deliver enough current for all three options, what would be the negative aspects of each option?

 

[FunkyB]

Also, is there any such thing as a physical difference between a 150:600 trafo and a 600:150 trafo, all things like brand, series, core, gauge etc. being equal?
Is a typical primary wound different from a secondary?

 

[joaquins]

When talking of multifilar windings pro and sec are exactly the same, all the 4 windings of this transformers (quadfilar) are the same, so there's no difference at all. There are other cases in multilayer windings where there are differences, but for this application you may want to use multifilar. Using 1:3 ratio and ±15V rails there is no much sense to use the second opamp, you will have 20 something Vpp, so you can get the +29dBu figure which is really nice, you won't have clipping problems with such output stage, you can even go higher with the rails without much problem, ±24V and 1:3 output transformer is way too much headroom, you won't need that.

By the way, at the C option you could use the center tap (if you have it) to go to the inverting input of the second opamp, not need for the resistors there, maybe there are some advantages doing so, I don't know. In any case, I don't see the point using the second opamp.

JS

 

[Monte McGuire]

Quadfilar output transformers generally seem to like being driven by a low source impedance, since their cores don't always include at least a few slices of pricey, high nickel content laminations. Driving such a typical grain oriented silicon steel transformer with a higher impedance will give you more core nonlinearities at low frequencies, and fortunately, this can be reduced simply by providing a low driving impedance.

However, some folks like this low end smoosh, and it's generally easy to accentuate by adding series resistance between the driver circuit and the primary. To reduce this, reduce the output impedance of the driver, and possibly arrange for a pair of drivers, one for each of the two primaries. What this will do is reduce the effective series resistance that drives each coil, since the two coils are no longer in series, reducing the driving impedance for each coil significantly.

You mentioned: "A plus could be that the output transistors would only have to deliver half the current." and this is indeed a benefit for some types of drivers, especially low feedback discrete designs. I worked on some older 800 series McCurdy modules that derived a second output for the other 150 ohm coil by simply tacking on another output transistor, paralleled to the main output device that was under overall loop feedback. So, the additional output transistor was running open loop, mimicking the output device that was under feedback. Sounds like a mess, but oddly enough, the second 'sidecar' output actually did the right thing at extremely low cost and undetectable extra distortion. Go figure...

Given the possibly random (high) load impedance on the secondary, it's probably a good idea to use a zobel on the secondary to prevent potential peaking. This will have to be determined empirically for all but the nicely specified Jensen transformers, but still, this peak will be way up there, especially if you have a low driving impedance.

Quadfilar output transformers seem to break a lot of rules, but in my experience, they're pretty serviceable and fairly malleable. The only thing I'd emphasize is that many of these devices use sorta scummy laminations that require very low driving impedances to get low LF distortion.

I'd also think of the device as four tightly coupled 150 ohm coils, and not 600:150, 150:600, 600:600 etc. etc. Any can be a primary, and any can be a secondary. These coils need to be driven with far lower than the 150 ohm spec to work halfway decently, so the whole impedance exercise is somewhat academic. But, by all means, play with it - there is a lot that one can do to tune the transformer by altering the driving circuit.

Best of luck!

 

[joaquins]

There are some topologies which cancels out the DC resistance of the primary, it's about 7Ω for the 2503, and the effective output impedance of the opamp could be considered zero for this analysis. Yet, the resistance at higher frequencies, if you analyze the total impedance at let's say 10Hz, you will find reactance of 257Ω and a resistance of 112Ω! That's due to core losses, so what's the point of canceling out the 7Ω of DCR I don't really know, I think you could go higher, taking into account the resistance at 20Hz for example and work over it but you would have to take into account the frequency response of that feedback or you will get stability issues due to that compensation at lower freq.

Maybe using a model with parallel resistance instead of series resistance for the core losses it makes more sense to compensate the 7Ω, which are effectively in series, and forget about the other, which you have to drive but it's not affecting the signal in the "ideal transformer" pri. I should think about this a little more, maybe some bench test would be good also, I wish I could do it right now, mind itching.

For your purposes, I think the Edcor should work fine, as I said, 1:3 configuration. The output stage of the API doesn't have much secrets, a cap in series, to get rid of the DC offset, which now is being compensated by a DC servo in the 512 for example but the cap is still there to keep the original sound, a resistor in parallel with that cap is there to avoid the LF peak of the resonance between the cap and the pri inductance, it does let pass a bit of DC but much lower current so there is no problem there, I don't remember the value now.

JS

 

[ricardo]

The only problem with this transformer is it is specified for a maximum voltage of 7.5VAC or just under +20dBu.

The max operating voltage level for a 'small' signal transformer is frequency & level dependent at LF.

This 7.5VAC spec is for a specified THD at some LF, probably 30Hz.  It is also dependent on the source & terminating impedances.

eg a proper spec would be 3% THD at 7.5VAC 30Hz, fed with 150R and terminated by 600R.

This would be 3% THD at 15VAC 60Hz, 1.5% THD at 3.75VAC 30Hz etc

And if driven by 75R, THD would be halved etc .. the limiting factor for THD reduction being DC resistance of the windings.

This is a far more accurate measure of the LF 'goodness' of a transformer cos it is precisely repeatable ... unlike measures of Inductance.

.. so what's the point of cancelling out the 7Ω of DCR I don't really know

Cancelling out the DCR drastically reduces this mostly 3rd harmonic distortion at LF.

There is a Lundahl application note for this.  We used a similar scheme at Calrec in the early 1980s.  Lundahl mention a German patent but I think Ken Farrar at Calrec developed this independently.

You can't reduce the DCR to zero cos there is a HUGE subsonic peak if you do.  Cancelling half is about right.

 

[joaquins]

...
e.g. a proper spec would be 3% THD at 7.5VAC 30Hz, fed with 150R and terminated by 600R.

This would be 3% THD at 15VAC 60Hz, 1.5% THD at 3.75VAC 30Hz etc
...

Distortion is generated by unlinearities, and they aren't linear, so you can't do that, if you have a HARD clipping, ideal linear response below that, for example and you are just below the saturation you will have 0% THD, if you just hit it a bit harder the distortion will start to raise, let's say 1V is out threshold, at 1.999V distortion will be really big, at half that will be zero, not the half of the first number. You are right it should be better specified, though.

You can't reduce the DCR to zero cos there is a HUGE subsonic peak if you do.  Cancelling half is about right.

That's why I mentioned the LF problems, what I'm saying is that if you put a HPF at the feedback, let's say at 15Hz and you compensate more than the DCR, or just that, you may not have the subsonic peak, since the compensation is not acting at that frequency, only where it matters for you.

JS

 

[ruffrecords]

But that's per winding, so with three in series as output it's about +29dBu, so pretty close to what you are probably looking with this transformer.

JS

Yes, you are right. Unfortunately I only read that part after posting.

Cheers

Ian

 

[ruffrecords]

The only problem with this transformer is it is specified for a maximum voltage of 7.5VAC or just under +20dBu.

The max operating voltage level for a 'small' signal transformer is frequency & level dependent at LF.

This 7.5VAC spec is for a specified THD at some LF, probably 30Hz.  It is also dependent on the source & terminating impedances.

eg a proper spec would be 3% THD at 7.5VAC 30Hz, fed with 150R and terminated by 600R.

This would be 3% THD at 15VAC 60Hz, 1.5% THD at 3.75VAC 30Hz etc

And if driven by 75R, THD would be halved etc .. the limiting factor for THD reduction being DC resistance of the windings.

This is a far more accurate measure of the LF 'goodness' of a transformer cos it is precisely repeatable ... unlike measures of Inductance.

Thanks for the insight. I certainly throws an interesting light on some tests I did recently with the EZTubeMixer output transformers. The tube line out stage, with its modest 20dB NFB cannot approach the near zero output impedance of an op amp. This means LF transformer distortion can be a serious issue at high levels. I recently compared a Carnhill VTB2291 wired for 2K4:600 (2:1) and an Edcor XSM2.4K/600 transformer driving +26dBu onto a 600 ohm load both driven by the tube line output stage. Despite the Carnhill having nearly twice the core size and twice the inductance of the Carnhill, in this circuit, their performance was near identical, both producing large amounts of LF distortion. Changing the load to10K drops the LF distortion close to mid band levels .

Cheers

Ian

 

[joaquins]

...
This means LF transformer distortion can be a serious issue at high levels. I recently compared a Carnhill VTB2291 wired for 2K4:600 (2:1) and an Edcor XSM2.4K/600 transformer driving +26dBu onto a 600 ohm load both driven by the tube line output stage. Despite the Carnhill having nearly twice the core size and twice the inductance of the Carnhill, in this circuit, their performance was near identical, both producing large amounts of LF distortion. Changing the load to10K drops the LF distortion close to mid band levels .

Cheers

Ian

That sound that the output stage is being overloaded rather than the transformer, if I'm right the transformer distortion should be lower with lower impedance loads, since the flux would be less, the clear example is the current mode or zero impedance input topologies, or whatever you want to call it (for input transformers). The input impedance is near zero and is used to need smaller sizes of cores with better LF response. The output stage is normal to have more distortion with lower impedance loads, so I think your distortion is coming from there, not the transformer it self,  the transformer might has something to do, but doesn't seems to be the source of your distortion by itself. Maybe I'm wrong or missing something.

JS

 

[ricardo]

e.g. a proper spec would be 3% THD at 7.5VAC 30Hz, fed with 150R and terminated by 600R.
This would be 3% THD at 15VAC 60Hz, 1.5% THD at 3.75VAC 30Hz etc

Distortion is generated by unlinearities, and they aren't linear, so you can't do that, if you have a HARD clipping, ideal linear response below that, for example and you are just below the saturation you will have 0% THD, if you just hit it a bit harder the distortion will start to raise, let's say 1V is out threshold, at 1.999V distortion will be really big, at half that will be zero, not the half of the first number.

The 'clipping' behaviour of 'ferrous' transformers at LF is well documented and very predictable.

Why don't you measure ANY small signal transformer for THD at LF at various levels and frequencies and tell us what you find?

 

[joaquins]

e.g. a proper spec would be 3% THD at 7.5VAC 30Hz, fed with 150R and terminated by 600R.
This would be 3% THD at 15VAC 60Hz, 1.5% THD at 3.75VAC 30Hz etc

Distortion is generated by unlinearities, and they aren't linear, so you can't do that, if you have a HARD clipping, ideal linear response below that, for example and you are just below the saturation you will have 0% THD, if you just hit it a bit harder the distortion will start to raise, let's say 1V is out threshold, at 1.999V distortion will be really big, at half that will be zero, not the half of the first number.

The 'clipping' behaviour of 'ferrous' transformers at LF is well documented and very predictable.

Why don't you measure ANY small signal transformer for THD at LF at various levels and frequencies and tell us what you find?

I don't have the rig here to do so, I will do it when I can, back at home, maybe in a month. I agree with you the proper way to do the data sheet is with freq, level and impedances, but still I don't think is as linear as you say after, half the level half the distortion.

JS

 

[ricardo]

The 'clipping' behaviour of 'ferrous' transformers at LF is well documented and very predictable.

Why don't you measure ANY small signal transformer for THD at LF at various levels and frequencies and tell us what you find?

..  but still I don't think is as linear as you say after, half the level half the distortion.

It's a low frequency effect and directly related to peak flux levels in the core.

From the specified eg 3% THD at 50 Hz 10V, THD is inversely proportional to frequency and proportional to level.

This works well from about 10% to 0.1% for 'ferrous' (including fancy iron) transformers and also mains transformers.  The THD is predominantly 3rd.

Below 0.1% THD, other effects start to dominate so this relationship doesn't work at Mid & HF unless you have a very small core.

There's probably stuff in Electronic Design Handbook 4 ed about this too.

 

[joaquins]

The 'clipping' behaviour of 'ferrous' transformers at LF is well documented and very predictable.

Why don't you measure ANY small signal transformer for THD at LF at various levels and frequencies and tell us what you find?

..  but still I don't think is as linear as you say after, half the level half the distortion.

It's a low frequency effect and directly related to peak flux levels in the core.

From the specified eg 3% THD at 50 Hz 10V, THD is inversely proportional to frequency and proportional to level.

This works well from about 10% to 0.1% for 'ferrous' (including fancy iron) transformers and also mains transformers.  The THD is predominantly 3rd.

Below 0.1% THD, other effects start to dominate so this relationship doesn't work at Mid & HF unless you have a very small core.

There's probably stuff in Electronic Design Handbook 4 ed about this too.

If you say so, I believe you, I didn't know about that linear relationship, thanks for the info, I'll look for it and run some experiments, it should happen in a transformer I wound myself like 2503 but trifler, right?

JS

 

[ruffrecords]

...
This means LF transformer distortion can be a serious issue at high levels. I recently compared a Carnhill VTB2291 wired for 2K4:600 (2:1) and an Edcor XSM2.4K/600 transformer driving +26dBu onto a 600 ohm load both driven by the tube line output stage. Despite the Carnhill having nearly twice the core size and twice the inductance of the Carnhill, in this circuit, their performance was near identical, both producing large amounts of LF distortion. Changing the load to10K drops the LF distortion close to mid band levels .

Cheers

Ian

That sound that the output stage is being overloaded rather than the transformer, if I'm right the transformer distortion should be lower with lower impedance loads, since the flux would be less, the clear example is the current mode or zero impedance input topologies, or whatever you want to call it (for input transformers). The input impedance is near zero and is used to need smaller sizes of cores with better LF response. The output stage is normal to have more distortion with lower impedance loads, so I think your distortion is coming from there, not the transformer it self,  the transformer might has something to do, but doesn't seems to be the source of your distortion by itself. Maybe I'm wrong or missing something.

JS

The test results seem to indicate the output stage is not clipping. Here is the response and distortion at 0dBu:

https://drive.google.com/file/d/0B_n67A1hN3qtVlNRZUdUbkQzNm8/edit?usp=sharing

and then at +20dBu:

https://drive.google.com/file/d/0B_n67A1hN3qtcm5BUzBSNFRESE0/edit?usp=sharing

both into a 600 ohm load.

Cheers

Ian