Correct output transformer termination for LL1517?

shop deals on amazon
Shop deals on ebay
Advertise with us
GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
T

tarandfeathers

Member
Joined
Apr 24, 2007
Messages
17
Location
Leeds, UK
Hi,
I recently built an 8 channel mic pre that uses Lundahl 1517s on the outputs. At present these are only terminated by whatever I plug the outputs into. On some of the channels I get a lot of interference (sounds a bit like AM radio static) on the last step of the gain switch. I'm wondering if this is some kind of ringing in the output transformer, as it seems to be significantly improved (though not completely remedied) by a makeshift RC termination (560R in series with 22nF). Most of Lundahls datasheets list a recommended termination but not for the 1517. Anyone know what the right values to use would be, and/or if doing so would be likely to fix the problem I've described?

Cheers,
Daniel.

Edit: I should add that the transformers are wired 1:2.
 
It is unlikely that the output transformer is the source of interference. It rather sounds to me as if you'd get close to oscillation at the highest gain position. This sort of behaviour is indeed not unknown to me and output transformer termination can have an effect on it, though it is propably not the main source of the problem. Does the sensitivity change with source impedance? In any case it will be helpful if you show the schematic and pictures of the wiring.

Samuel
 
Hi,

Unfortunately I don't really have a sensible way of getting photos onto my computer at the moment. The preamps are based on the Focusrite 110 design, with a couple of minor adjustments (1576 for the input transformer rather than the 1538, and the 1517 on the output). I'd post the schematic but I know Focusrite are notoriously funny about people putting their stuff on the web. It's basically the input transformer followed by a stepped attenuator, a 5534 with stepped gain and then straight into the output transformer.

You're right enough about the oscillation, looking on the oscilloscope on the last step of gain theres a signal somewhere around 67 kHz which is about 55 V peak to peak. What kind of things cause this type of oscillation? It seems odd that it does it on some channels and not others. The reason I suspected the output transformers was because I previously had a 4 channel unit that was identical except it had unbalanced outputs, and that didn't display this problem.

I'll try and sort some photos of the inside of the case, if that'll help.

Thanks,

Daniel.
 
The preamps are based on the Focusrite 110 design.
Click to expand...
So it looks about like that one: F110_schem2.PDF

You're right enough about the oscillation, looking on the oscilloscope on the last step of gain theres a signal somewhere around 67 kHz which is about 55 V peak to peak.
Click to expand...
Are you sure about the numbers? 55 Vpp is surely too much from a poor opamp running on +/- 18 V supplies. And 67 kHz is very low for true oscillation. 5.5 Vpp and 670 kHz would make sense.

What kind of things cause this type of oscillation? It seems odd that it does it on some channels and not others. The reason I suspected the output transformers was because I previously had a 4 channel unit that was identical except it had unbalanced outputs, and that didn't display this problem.
Click to expand...
It's similar to the feeback you get when mixing live gigs--instead of the loudspeaker feeding it's signal to the mic it's the output feeding it to the input. If the output (or input) transformer got an amplitude peak at high frequencies (which might vary from sample to sample) then the gain at this frequency is increased and hence the chance of oscillation greater. A zobel network will dampen the response and work against oscillation.

To solve the problem I suggest the following approach:
a) Check the wiring. Distance between output and input greatly reduces capacitive coupling. Don't forget the gain switch and stepped attenuator.
b) Check the IC bypass capacitors. 100 nF across the supply pins within half an inch of the opamp is a must.
c) Verify the zobel network of the input transformer. You've used a different transformer, so the zobel will need different values as well.
d) Add a zobel to the output transformer.

By now the oscillation should have disappeared I'd say. To verify try various source impedances from 0 ohm (short between pin 2 and 3) over standard 150 ohm to open. I suggest you carry out all four steps even if an early one solves the problem because that's just good design practice.

If not, you might try:
* to add shielding. Use shielded cables for the XLR connections and perhaps to the gain switch and attenuator as well. And remeber that only a grounded shield is a good shield.
* a different grounding scheme (if it's not a PCB).
* to reduce bandwidth of the opamp gain stage(s) by increasing the feedback capacitor.

Just to make sure: you didn't try to include the output transformer into the output driver feedback loop as in the original ISA110? That won't work well without the original transformer.

To determine the zobel networks: apply a fast square wave (10 kHz to 30 kHz for a transformer) to the input (at low gain) and search the highest-value resistor which will give little or now overshoot (use a 10x probe only). Now add the smalest capacitor value which still gives little overshoot. You might need to adjust the resistor value a bit for best result.

For the output zobel I might suggest that you temporarily bypass the input transformer (and perhaps the first gain stage as well) to avoid early bandwidth-limiting.

Samuel
 
[quote author="Samuel Groner"]So it looks about like that one: F110_schem2.PDF[/quote]

Yeah, that's nearly it. I have an original Focusrite schematic, and it's all the same except R10 is 220k...? I tried changing this to 22k and the oscillation is still there.

Are you sure about the numbers? 55 Vpp is surely too much from a poor opamp running on +/- 18 V supplies. And 67 kHz is very low for true oscillation. 5.5 Vpp and 670 kHz would make sense.
Click to expand...

I double checked, that's definitely what I have. The 55V is after the output transformer though, so there's 1:2 step up on that.

To solve the problem I suggest the following approach:
a) Check the wiring. Distance between output and input greatly reduces capacitive coupling. Don't forget the gain switch and stepped attenuator.
b) Check the IC bypass capacitors. 100 nF across the supply pins within half an inch of the opamp is a must.
c) Verify the zobel network of the input transformer. You've used a different transformer, so the zobel will need different values as well.
d) Add a zobel to the output transformer.
Click to expand...

Ok,
a) The input and output leave the board right next to each other, and the input and output transformers are close to each other on the pcb, but since the problem isn't on every channel maybe the distances are ok? I hope so, or else I'll have to redo my boards...
b) This is fine.
c) I increased the resistor in the network to 22k, but perhaps this isn't high enough. I will try some higher values.
d) This is next on the list, along with C.

Just to make sure: you didn't try to include the output transformer into the output driver feedback loop as in the original ISA110? That won't work well without the original transformer.
Click to expand...

No, I didn't do that. Well, I'll try improving the zobel networks, and if that doesn't work, I'll change the feedback cap. The input and output wiring is already done with sheilded install cable, I might try sheilding the cables to the switches if the above doesn't work.

I tried changing the source impedance: when pins 2 and 3 are shorted the signal is much lower, but still present, lower still at 120R and gone by 270R. It is gone for all source impedances up to 1k2, but back by 2k2, which is the next highest value I had handy.

Thanks again for all your help!

Daniel.
 
Hey tarnand, Look at your supply +/-feeds to those 8 boards and of course your grounding within the chassies. Try to feed every board with a series resistor 10-22 ohm and a cap 470- uF to ground. Stargrounding is also a need. Ive used the the 1576 with just 68k across the sec. -no ringing.Look at the schem. the 22pF for the 5534 is to go from pin 5 to 8. Cheers Bo
 
Hi Bo,
Everything you've mentioned is correct. The preamps are on two boards of four but each one is a completely seperate circuit with independent connections to the power supply, and each IC has in-line resistors and decoupling caps. The compensation cap is also already in the right place. I'll have a look at fixing the transformer terminations in the next couple of days, I suspect that'll sort my problem out.
Cheers,
Daniel.
 
I tried changing the source impedance: when pins 2 and 3 are shorted the signal is much lower, but still present, lower still at 120R and gone by 270R. It is gone for all source impedances up to 1k2, but back by 2k2, which is the next highest value I had handy.
Click to expand...
That's a pretty good indication that you need to check the zobel of the input transformer. Higher source impedances provide more damping, that's why the oscillation dies. Why it get's back again with higher impedances is not clear to me, perhaps the input get's more sensitive to capacitive coupling again.

To provide equal damping with lower source impedance you'll need lower resistance and probably more capacity.

I increased the resistor in the network to 22k, but perhaps this isn't high enough. I will try some higher values.
Click to expand...
As I said above: lower values, not higher ones!

My description on finding zobel values was no perfectly clear. You should scope the square wave right at the secondary of the according transformer, not after the active stages.

BTW, what's your maximum gain? If the input transformer has higher step-up than the original part and the output transformer adds some 6 dB you might get above 70 dB which will not simplify things. Just use as much gain as you'll need, not more.

Oh, and another BTW: how does the output stage look like? A NE5532/NE5534 might become relatively unhappy while driving a 1:2 transformer if there's no additional output buffer (of whatever form) helping to provide the required current.

I've used the the 1576 with just 68k across the sec.--no ringing.
Click to expand...
Sure? I've done so as well and there was lots of (at least 6 dB peak at ~150 kHz IIRC)... Did you check right at the secondary? What source impedance and frequency?

Samuel
 
Hey, those +6 dB at 150kHz wasnot any problem...to answer Samuel, gen imp. 200ohms square wave at 10 kHz at diff level. The amp stage is a clone to Tube-Tech MP first stage followed by a fet outp stage.Back to..
In my 30 years experience of preamps I claim that those most well sounding are not those with highest bandwith.My 4 sets of IRT V76 are my favourite pre and after that the Forsell Fet-amp,the Neve 1272 and so on.This is clerly subjective but has BW in common.
My believe is to roll of at about 85kHz,not to sharp,6dB/octave.Of course, if you spot a ringing problem take it away otherwise you gonna hear it.

tarand,what gen imp have you calculated? The resist or sec pot cant be as low as the ISA 110 org with the Tx in 1:7 mode.Is there two amps in your pre channel and if so what kind of connection is there between those. If not I think your single 5534 gets stressed with that 1:2 loading.Is it AC or DC conn. to the 1517?
Cheers Bo
 
[quote author="bovox"]tarand,what gen imp have you calculated? The resist or sec pot cant be as low as the ISA 110 org with the Tx in 1:7 mode.Is there two amps in your pre channel and if so what kind of connection is there between those. If not I think your single 5534 gets stressed with that 1:2 loading.Is it AC or DC conn. to the 1517?
Cheers Bo[/quote]

By 'gen imp' do you mean source impedance? The source feeding the input transformer, or the impedance that the input of the op amp sees? I only have the single 5534 in each channel, coupled to the output transformer through a 100uF capacitor to block DC. Before I started on this 8 channel version I tried putting a 1517 onto the output of my previous four channel one to see if the amps would handle that and they were fine.

And yeah, the gain is pretty high, certainly over 60dB, I've never yet needed that last step of gain, but I want to make sure it works just in case....
 
Hey tarand, If we want the mic to see about 1kohms at the input then in 1:7 mode the Tx outp imp is about 50kohms. If you take the org res.ladder around 33 kohm, then your mic sees around 600 ohms.
With the 100uF at the output and if the Tx is loaded with 600ohms you get a roll of with 6 dB at 10 Hz...take a 470uF or bigger cap.
I also think that you should have a separate buffer for that out Tx working in 1:2 mode. Cheers Bo
 
Hi there!.
I'm new to this furom, though I've been reading it for some times, this will be my first post.

I greatly apreciate this discution as it answers questions about rining, osclillatin opamps as well as loading and Zobel value selection.

Does anyone can redirect me on leactures going deeper on those subjects, Detecting rigning, xfo impedance etc.

Hopin' to find an aswer...

Jo Many thanks to all.
 
Samuel Groner said:
The preamps are based on the Focusrite 110 design.
Click to expand...
So it looks about like that one: F110_schem2.PDF

You're right enough about the oscillation, looking on the oscilloscope on the last step of gain theres a signal somewhere around 67 kHz which is about 55 V peak to peak.
Click to expand...
Are you sure about the numbers? 55 Vpp is surely too much from a poor opamp running on +/- 18 V supplies. And 67 kHz is very low for true oscillation. 5.5 Vpp and 670 kHz would make sense.

What kind of things cause this type of oscillation? It seems odd that it does it on some channels and not others. The reason I suspected the output transformers was because I previously had a 4 channel unit that was identical except it had unbalanced outputs, and that didn't display this problem.
Click to expand...
It's similar to the feeback you get when mixing live gigs--instead of the loudspeaker feeding it's signal to the mic it's the output feeding it to the input. If the output (or input) transformer got an amplitude peak at high frequencies (which might vary from sample to sample) then the gain at this frequency is increased and hence the chance of oscillation greater. A zobel network will dampen the response and work against oscillation.

To solve the problem I suggest the following approach:
a) Check the wiring. Distance between output and input greatly reduces capacitive coupling. Don't forget the gain switch and stepped attenuator.
b) Check the IC bypass capacitors. 100 nF across the supply pins within half an inch of the opamp is a must.
c) Verify the zobel network of the input transformer. You've used a different transformer, so the zobel will need different values as well.
d) Add a zobel to the output transformer.

By now the oscillation should have disappeared I'd say. To verify try various source impedances from 0 ohm (short between pin 2 and 3) over standard 150 ohm to open. I suggest you carry out all four steps even if an early one solves the problem because that's just good design practice.

If not, you might try:
* to add shielding. Use shielded cables for the XLR connections and perhaps to the gain switch and attenuator as well. And remeber that only a grounded shield is a good shield.
* a different grounding scheme (if it's not a PCB).
* to reduce bandwidth of the opamp gain stage(s) by increasing the feedback capacitor.

Just to make sure: you didn't try to include the output transformer into the output driver feedback loop as in the original ISA110? That won't work well without the original transformer.

To determine the zobel networks: apply a fast square wave (10 kHz to 30 kHz for a transformer) to the input (at low gain) and search the highest-value resistor which will give little or now overshoot (use a 10x probe only). Now add the smalest capacitor value which still gives little overshoot. You might need to adjust the resistor value a bit for best result.

For the output zobel I might suggest that you temporarily bypass the input transformer (and perhaps the first gain stage as well) to avoid early bandwidth-limiting.

Samuel
Click to expand...
Do you have the 110 output stage in schematic form too? The feedback loop that is.
 

Similar threads

trobbins
Vintage Mic Transformer for MC step-up - gain-phase and impedance measurements
5 6 7
Replies
130
Views
7K
Cqwet Dbdfte
Cqwet Dbdfte
B
Output transformers near mic level wiring?
Replies
8
Views
1K
bmaughan
B
B
Help choosing transformer for opamp output stage
Replies
5
Views
3K
Newmarket
Newmarket
rackmonkey
Signal Amps for output transformer testing
Replies
5
Views
4K
alexc
alexc
M
Power supplies and smaller p2p boards for 2ch tube preamp
2
Replies
32
Views
9K
PRR
PRR

Latest posts

Back
Top