LA2A with Edcor Transformers - PROBLEMS!!!

[khstudio]

[quote author="emrr"]An obvious thing that seems to have eluded us all. You are using the XSM series, which have worse response specs than the WSM series. I doubt you need the level handling ability of the XSM; even the WSM stuff has around 10db more headroom then the original UTC iron. Your response measurements with the XSM input are in spec. I'd try the WSM since you've already done so many comparisons.[/quote]

I Know... the only reason I bought these to use was from other guys using them on "similar" builds. + there cheap.
The input especially is VERY BIG compared to what is needed.
Right now I'm just running Un-Balanced input until I get my CINEMAG or JENSEN's in

I paralleled a 16uf across the 10uf to get a total of 26uf on the output CAP.
This brought that bump in the lows down.

UnBal-in - 26uf output Cap (C3) EDCORE XSM 10K:600- NO LOADING

UnBal-in - 26uf output Cap (C3) EDCORE XSM 10K:600- 600 Ohm LOAD!

UnBal-in - 26uf output Cap (C3) EDCORE XSM 10K:600- 600ohm Sec. & 11k LOAD Primary

 

[emrr]

cap change helping for sure. I'm gonna speculate wildly and suggest a 27K primary load resistor might get you closer to flat. Call it an illiterate gut feeling. Reality check of course being that we're close enough already and are splitting tiny hairs for the sake of honing our blades.

 

[khstudio]

[quote author="Gus"]What are you using to drive the input transformer?

What is the output resistance of the signal source?[/quote]

Hey Gus,
How you been?
I got you message but I'm going out to dinner.
Thanks for helping out :thumb: I'll give you a call later on.

About the driving & impedance... if you look earlier in the thread, I tested the input TX OUT of the case... loaded & un-Loaded & it didn't have any of the high shelf looking roll-off. Must be something in the circuit.
Other trannys or running un-balanced in, don't do it either.
I'm done with these on the input.
They're TOO BIG, Un-shielded & not working... not worth any more waisted time.
$70 for the Jensen 10k:10k
$74 for the CINEMAG 10k:15k
seems like the way to go... a know good tranny for input.

The EDCORE is working for the output but I still may replace it & just use them for some tube Pre's I plan on building, like the MILA.

 

[khstudio]

[quote author="emrr"]cap change helping for sure. I'm gonna speculate wildly and suggest a 27K primary load resistor might get you closer to flat. Call it an illiterate gut feeling. Reality check of course being that we're close enough already and are splitting tiny hairs for the sake of honing our blades.[/quote]

Hey Doug,
I'll try that later when I get home.

I wish I realized the output cap thing BEFORE I bought that Mallory TC72 10uf/450v (Original cap) :sad:

Although... when I was looking thru my parts bin & found a Mallory TC75... which is a 20uf/450v & it measured perfect :green:

All this output cap stuff makes sense now.
JENSEN sent me a 22uf/450v cap with their transformer :!:
I installed it & never looked back.
BUT I didn't use ANY LOADING with the Jensen & as you can see, it measures & looks great.

 

[emrr]

Loading is not a hard and fast thing, as we see by this. Some transformers will behave differently from others. I've seen 10K:10K inputs loaded with 10K on the sec that only look correct with low impedance (100-200 ohms drive) source, and others that look fine with a 10K source. I'd still get a WSM series input and see how it does.

 

[khstudio]

The Cinemag's I ordered are:
Input = CMLI-15/15B 15k:15k $45

Output = CM-9589 10K : 600 $74

I'm done with the EDCORs on my LA2A even though the output works OK.
I'll be using the OT's on something else.

 

[SSLtech]

Kev,

I ran sweeps and measurements today. I calibrated R25 (meter tracking) and a couple of other tweaks. I got the response to within 0.2dB 20Hz-20kHz.

I swear that it's probably not an Edcor-specific issue: that circuit might not like the impedances which you've chosen to use... The WSM 600:15K works excellently on the input (15R source Z) and the WSM 15K:600 seems to be about as good as the UTC A-24 on the output.

Oh yeah... and distortion was down around 0.075% when I'd finished tweaking... WITH the T4 fitted...

Keith

 

[khstudio]

[quote author="SSLtech"]Kev,

I ran sweeps and measurements today. I calibrated R25 (meter tracking) and a couple of other tweaks. I got the response to within 0.2dB 20Hz-20kHz.

I swear that it's probably not an Edcor-specific issue: that circuit might not like the impedances which you've chosen to use... The WSM 600:15K works excellently on the input (15R source Z) and the WSM 15K:600 seems to be about as good as the UTC A-24 on the output.

Oh yeah... and distortion was down around 0.075% when I'd finished tweaking... WITH the T4 fitted...

Keith[/quote]

That's good to know. :thumb:


Well, my LA2A with Jensens 10k:10k (with the same 11k secondary load) looks GREAT... so I'm not sure about the circuit loading being a problem.
&
the Edcor 10k:10k input, outside the circuit, DOES looks good.
&
The LA2A Un-Bal with a 10k load looks good.

Is just strange when they're put together.
Instead of Peanut Butter & Jelly... it's like Jelly & Anchovy's :shock:
... I kid I kid. :green:

Once again... I do like the output TX.
When I compare this new one & my older one (with Jensen's) the Jensen one beats it ALL the way around... it just sounds "Perfect" :green:

BTW, The Edcor LA2A vers. does NOT suck... just not "as" good.
So I figured, $125 for the set of CIMEMAG's is not too much to pay for what I'm looking for + I spent way too much time already.
I've got an 8ch API on the bench & a stereo Pultec. :green:

I trust you Keith & appreciate your input.
I just wanted to share what I found with these models & the LA2A.

 

[khstudio]

Keith,
Maybe it's also the XSM's as opposed to the WSM series?
The input is WAY LARGE & Heavy... probably too much.

Doug mentioned the WSM series being specked better & I should try those.
But I've already done my share of testing Edcors (for now) for this project.

The ONLY reason I got the XSM's was from reading others post about them & recommend them. Not sure if they actually used them & or tested them with a sweep or any other equipment.

I could & would, live with the OUTPUT TX... but I'm gonna use them on my NYD MILA outputs.

 

[Svart]

I haven't been doing much DIY these days and subsequently haven't been around here much but I've brought the LA2a back out and I'm battling with the edcors on the input as well. 600:10k on input gets me horrible oscillations. They are there no matter what kind of loads, impedance matching, gain, or zobel networks I add. At the voltage divider before the T4b, I see a 5mv complex waveform in the noise. It's pretty small but comes out of the noise at the node between C1 and R12, after the first gain stage. After the second stage, measured at C2, the wave looks like a sinusoid with folded peaks. At this point it changes with gain so must be coming from the input. My power rails are clean with less than 10mv ripple after I added some more decoupling to the circuits as well as using soft/fast rectifiers.

I removed the transformer and the waveform goes away even without input termination, or input signal for that matter. When I input an output from a preamp I get horrible hum AND hiss as well as a severely distorted sound but when I input a generated signal I get clean waveforms through the whole unit all the way down to a 100mv input. So as it stands I don't see the distortion that is present when a preamp (1272) is the input source on a lab setup.

I have tried a number of different grounding schemes as well as the above mentioned impedance matching and nothing gets rid of this problem. My other transformers are in storage so I cannot test anything else right now. I swept the edcor xfrmr and it looks great by itself.

These are also Xsm parts and I am using the Drip version 10 board, from the first group buy..

Any thoughts?

EDIT: Keef, is that an FD4 in your avatar?

 

[guitarbill55]

FWIW I used a Cinemag CM9589 (HiNi) version output on mine and I love it. I also used the Cinemag CMMI-7C ( HiNi ) version 50/200 to ? 10K and I think I love that input, too, though at that ratio I have to watch the input levels. Mine are all point to point ala caya, but I think you'll like the Cinemags. BTW I didn't use any variable or fixed roll off caps in mine and I still like the way they sound and behave.

gb

 

[Svart]

The original was 600:60k. don't think I'll have to worry about 200:10k.

I would like to understand why these transformers are not working more than just getting it to work at this point.

 

[rodabod]

[quote author="Svart"]Keef, is that an FD4 in your avatar?[/quote]

I've been thinking the same. I hated that thing.

 

[guitarbill55]

Ooops, Svart, my bad! The Cinemag CMMI-7C is spec'd for a secondary load of 97 k ohms! It's a 1/7 turns ratio. That's why it seems so hot going in. I did put a question mark in the first post for a reason: It's tough to get old. Sorry for the confusion. I shoulda looked at the spec sheet.

gb

 

[SSLtech]

Again, I've just fitted WSM 600:15Ks in and out on another LA-2a, and the results are excellent.

Can I suggest the WSMs, -Svart, I didn't see which series you were using?

-As for the gear in the picture, I don't remember what they gave me to hold... -they just shoved some porta-studio type thing in my hands and told me to grimace!

Keith

 

[Svart]

Yup that's an FD4.

I have one in the basement that I used once..


http://www.edcorusa.com/products/transformers/xsm/xsm600-10k.html

those are the transformers I am using.

2.5W compared to .5W for the WSM series. I expect the windings to be thicker. I suppose inductance is playing a role here.. I suppose I could have them wind me a 600:60k WSM series and give it a go.. Only 10$ to waste..

 

[emrr]

sigh.

I'm gonna stop saying it. I second Keith's comments.

The A-10 and A-24 are rated at +15dbm.

0.5W WSM series Edcor = +27dbm = already way more than an LA-2A puts out in the first place.

10K:10K Jensen/Edcor etc = -1 db or less transformer insertion loss
600:10K Sowter/Edcor = +/- 13 db more gain
600:50K UTC/etc = +/- 21 db more gain


You get 7 db more gain with the stock 50K winding, which almost no one needs, and unless wound correctly, a good chance of high frequency loss from rising winding capacitance. If you want a 50K secondary, getting Edcor to wind it is probably not the way to go. They may even refuse the job.

 

[Svart]

when I have to max out my preamp to get a little compression from a non-amped bass guitar while using a 600:10k transformer on the front end, that tells me that I need more gain on the front end. There was a reason that the HAX input tranny had 600:60k in the first place. I've run into this problem with really low input signals on the LA2A and wish to rectify it in a way that I will be able to use every single preamp I have regardless of max gain. I have some preamps that only do 40db and need that voltage gain from the front end of the LA2A. The easiest way to do this is clearly to change the transformer. I stuck the 600:10k in place since it was the only one I had available to use with something other than 1:1. Now I am wondering why the edcors seem to have trouble when others do not, yet the edcors are pretty good in other applications and on the bench.

I still don't see the mechanism for the transformer to self resonate with the inputs clamped to ground or otherwise terminated and why a completely open input to the circuit(no transformer or termination) is as quiet as a graveyard.

 

[SSLtech]

The Jensen transformers which I recently removed from this LA-2a (on my bench right at the moment) were doing exactly what you describe, Svart. -The Edcor WSM600:15K at the front end is the PERFECT solution. -MUCH more like the A-10 than the Jensen EVER was.

I also have a set of UA HA-100X and A-10 here, and I had them on the Bloo which I recently sold on eBay, which also sold with the Edcors 600/15k and 15k/600.

I have NO compunction whatsoever recommending that EXACT pair for the LA-2a: however, if anyone wants to try different transformers, -whether they be Edcors or some other brand- we've tried a LOT of different ones in the LA-2a's over the years, and I personally think that straying too far from the original set of ratios really does make it far too difficult to really 'dig in' to signals unless they're VERY hot.

Keith

 

[Silvas]

Has anyone tested the XSM600:15K on a working unit?