Input transformer query

[Thermionic Idler]

In the middle of building up the second of the 300B push-pull monoblocks, and I come across an Audio Asylum post regarding the Lundahl LL1684 input transformer I chose for the project. It's a small amorphous core low impedance transformer, that I selected to match the low-Z output of my preamp (66 ohms balanced).

It turns out that this particular transformer doesn't like DC offset very much - quote from Kevin C: "If there are even a few millivolts, this type of transformer is doomed to failure due to core saturation." After reading that, I remembered with a sinking heart that my linestage is a direct-coupled op-amp based affair with no coupling caps, although it is designed to minimize DC offset as much as possible.

So today I measured the DC between + and - phase on the pre-amp outputs. On the right channel it was 0.7mV, on the left it was 1.2mV, both figures are well within design tolerance. When I listened to the first completed amplifier build, it was on the right channel, so the IPTX was seeing the lower 0.7mV, and I couldn't hear any ill effects, it sounded glorious.

However, my question is what should I be listening for (or measuring for) to identify whether the slightly higher 1.2mV of the other channel is enough to tip the core into saturation? It's less than Kevin's "few millivolts" but there's no documentation specifically stating what kind of DC tolerance exists.

The series resistance of each primary is 41 ohms, and I have the primaries series connected (2:1+1), so they are presenting 82 ohms. 1.2mV would therefore result in a DC current of 14.63 micro-amps. Which really doesn't sound like a lot to me, so I'm just wondering if I'm worrying over nothing. I really don't want to have to swap the transformers out as they were not exactly cheap. I guess my other option is try to find out what can be done to reduce the DC coming out of the linestage.

Gah! You try to think of everything and spec your parts appropriately, and something always comes out of the woodwork to bite you.

Thanks in advance.

[IslandPink]

It doesn't sound like much. I think the bass would be affected first.
It should really quantified, so people know what they're working with.

[Nick]

Well, they are amorphous core not nickel so I would have expected they could stand a bit of DC flux. But yes, its the bass that would suffer. May be worth measuring the DC offset with the transformer connected, it may be less (if you measured the preamp open circuit). You can try a cap in series with the input and see if you hear any difference.

[Thermionic Idler]

It doesn't sound like much. I think the bass would be affected first.
It should really quantified, so people know what they're working with.

Exactly, it would have been nice if this had been documented on the data sheet instead of buried on the internet for me to find after I'd spent the money. It's all very well saying "zero DC" - in the real world how do you get to that without coupling caps in circuit.

@Nick - thanks for the help, I'll do some tests with the preamp in and out of circuit to determine if there's any issue.

[izzy wizzy]

I noticed you say the input is connected in series but the datasheet shows parallel connection for phase splitting. I tried series input connection on my 1676, and the phase splitting performance was all over the place between the two outputs compared to driving the input coils in parallel.

You could check the frequency response of the amp with the line connected or not. Adding a cap depending on value might cause a resonance boost in the bottom end. Some of the lundahls are quite well behaved but it will happen a bit. Just saying as adding a cap might lead to an incorrect conclusion.

Edit: the best phase splitting connections for the 1676 aren't on the datasheet either but buried on the net too.

[Max N]

The rakkdac I used to have used lundahl amorphous output transformers. In that case, Kevin/Dave put the parafeed cap between the two primary windings. You could fit one with a switch to short it out, and see if you can hear a difference. Assumes you are series connecting the primary.

[Max N]

On second thoughts, for ease of testing, a cap between preamp and amp would be much simpler

[Nick]

in the real world how do you get to that without coupling caps in circuit.

Transformer output. Just saying.

[Max N]

I'm in a similar situation with the Sowters I'm using. Sowter state 'no uncancelled dc permitted in primary or secondary..'. I'm using them with a dc-coupled dac, and measuring a few micro-amps. Bass sounds good (any concerns I have are at the other end)

Nick did suggest a while ago in the measuring transformers thread that I should check the effect of dc on the bifilar ITs, and also has suggested I try parafeed with the Sowters, so I am now belatedly getting set up to measure the effect on both and will post the results. Regardless of the measurement results, I will also try listening to the effect of a parafeed arrangement on both transformers. Work has got busy again so may be a slow process.

A question for the group - if I send dc current through a 100% mumetal laminated core, is there a danger of permanent magnetisation?

[Thermionic Idler]

I noticed you say the input is connected in series but the datasheet shows parallel connection for phase splitting. I tried series input connection on my 1676, and the phase splitting performance was all over the place between the two outputs compared to driving the input coils in parallel.

Interesting - I went for series connection because I had plenty of headroom in the preamp, and I figured I'd get better performance. When I built the last amp, after installing the input transformer with its load network, I scoped both phases on the secondary to make sure they were equal, and all looked fine. I'll be running the same test today for channel B.

I read somewhere that amorphous core was actually less tolerant of DC than other cores, but I trust Nick more than some rando on the internet.

Given the preamp output is perfectly balanced, I'm starting to wonder if I shouldn't just try direct coupling it. I did have a concern over whether that would create a ground loop because the power amps are plugged into different sockets near the speakers, but looking at it again I can't see a path that ground current could take, if I connect + and - phase floating - kind of the point of two phase output, not needing to rely on ground as a reference. I'd also lose RFI cancelling but a ferrite bead could probably take care of that.

[Nick]

Well yes amorphous will saturate before m4 but I would worry more if it was nickle. As to the balanced input. If you are going to valve grids then it won't handle a DC offset like a transformer input. It will cancel any ac difference up to a point but it won't be perfect so YMMV. Ok I would have a ground and see how it goes.

[izzy wizzy]

Interesting - I went for series connection because I had plenty of headroom in the preamp, and I figured I'd get better performance. When I built the last amp, after installing the input transformer with its load network, I scoped both phases on the secondary to make sure they were equal, and all looked fine. I'll be running the same test today for channel B.

It was your measurements that got me to measure mine. Easiest is to x-y the two phases to produce a lissajous figure. It should be a 45 deg line that doesn't change shape or length. In the end, mine only started to open a fraction above 50kHz.

Sorry of this is an eggs/grandma post.

[Thermionic Idler]

Interesting - I went for series connection because I had plenty of headroom in the preamp, and I figured I'd get better performance. When I built the last amp, after installing the input transformer with its load network, I scoped both phases on the secondary to make sure they were equal, and all looked fine. I'll be running the same test today for channel B.

It was your measurements that got me to measure mine. Easiest is to x-y the two phases to produce a lissajous figure. It should be a 45 deg line that doesn't change shape or length. In the end, mine only started to open a fraction above 50kHz.

Sorry of this is an eggs/grandma post.

Far from it - I didn't do that . I just overlaid the sine waves to see if they looked the same.

[Thermionic Idler]

OK I did some testing with the multimeter and the Analog Discovery - this is really quite a nice bit of kit. Turns out that I could simulate a DC offset on the waveform generator, although at the millivolt level it wasn't all that accurate - the AD itself seems to have a bit of an offset on its output that varies between 0.5 and 2mV depending on what the software is doing.

Anyway, I placed the multimeter across the input on the mV setting to give me the DC offset. I had to use a frequency of 1kHz as any lower than that resulted in the multimeter "reading" the signal as a value, even in DC mode. I set the input to quite a healthy 2V - I *think* that's a level that would take the amp past its clipping point.

So here's a screen grab of the Waveform software, showing offset set to zero in software and the response of the two secondary phases at very close to 1V:



With this setting, the multimeter read 0.8mV offset, same as the left channel preamp offset which I've already verified sounds fine:



I then set the offset value in software to 1mV, and there was no change in the amplitude or shape of the secondary waveform:



The multimeter was reading 1.9mV of DC which is quite a bit higher than what's coming out of the preamp on the right channel:



I then dropped the frequency to 20Hz at the same amplitude, and the secondary response remained unchanged:



That seems to indicate that I probably don't have anything to worry about, although I'm willing to bet someone will come along and point out something I've missed!

[Nick]

All I will point out is I would not look at sine waves, I would meeasure a frequency sweep if it can do that or a number of spot frequencys if not. And I would look at a FFT of the sine wave, if you can see distortion its up close to 1%