SOUND ON SOUND

[Hugh Robjohns]

I have found a schematic (not block diagram) for the CB2...

Oh. Well done. Please share!

...and there is Zobel CR network across the output.

I suspected as much! Thanks for comfirming.

Therefore using it 'the wrong way' is likely to throw the response all over the shop depending upon source and sink impedances as you say Hugh.

Yep! ;-)

[DanDan]

I have found it to be pretty useless when trying to cure a hum/buzz problem between a projector and PA system. Can't remember exactly but I think the circuit diagram shows the culprit lack of earthing.

[Martin Walker]

Like Hugh, I'm looking forward to viewing the CB2 schematic with Zobel network too :thumbup:

Forty years ago I saved the impedance matching transformer from an AKG drum mic (I think it was a D12, and was mounted inside an overlarge 1/4-inch jack plug), that converted this low impedance microphone so it could be plugged into a high impedance PA input.

I eventually built it into a DIY DI box 'the other way round' so that it ends up with a high impedance line-level input and low impedance output suitable for plugging into a mic input. Coming from AKG I assumed it was a quality item, and it always sounded good to me as DI box.

However, I never considered a Zobel network, but am beginning to wonder whether I should, particularly as I'm running it 'the wrong way round'. A quick check on its frequency response using the PluginDoctor utility (Dirac Pulse test signal) suggests that this rises significantly above 1kHz.

Can't find any online data on the transformer itself, but it's marked as AKG UT 330 MATCHING TRANSFORMER, which I suspect will have a 1:14 turns ratio, so with my RME Fireface UC mic input impedance of 2K the input impedance of my DI box ought to be around 392K (2k*14*14)

When I get the chance I'll check out a selection of R+C options across the secondary, to flatten out this response as much as I can - I was surprised just how non-linear it was! Perhaps it would be easier to send in a 1kHz square wave and tweak values for the 'squarest' shape with no overshoot?

I'm now hoping I can drive this hard to achieve some interesting 3rd harmonic colour ;)


Martin

[ef37a]

https://artproaudio.com/framework/uploa ... om_dti.pdf

Martin, taken me a heck of a time to find that link again! Unfortunately no CR values but Self gives a 1:10 transformer mic pre into an op amp with a 100k load and 220pf in series with 22k.

You could start with that then check the response at 20kHz? I would not use a square wave, probably ring like billyo!

However, you are using the transformer as a DI box? What are you driving it with? If guitar as is usual then I doubt the response will still be flat compared to that found on the 'test rig'?


Dave.

[Hugh Robjohns]

I have found it to be pretty useless when trying to cure a hum/buzz problem between a projector and PA system. Can't remember exactly but I think the circuit diagram shows the culprit lack of earthing.

:?: :think:

[Hugh Robjohns]

A quick check on its frequency response using the PluginDoctor utility (Dirac Pulse test signal) suggests that this rises significantly above 1kHz.

Yes, that's quite common.

My AP test system defaults to an input impedance of 200k Ohms, and that's so high as to be effectively an open-circuit load for a lot of transformer outputs. The result is almost always a high resonant peak around 50kHz or so.

Load the output with 10k Ohms and the response usually flattens out... and often loading with 600 Ohms introduces a gentle HF roll-off.

So when you come to experiment with your own taming Zobel network, make sure you also load the output to match whatever it normally plugs into, otherwise you might find you over-correct!

[Hugh Robjohns]

https://artproaudio.com/framework/uploads/2018/06/om_dti.pdf

Ah... yes. Familiar with that one. It's a bit of an over-simplification I fear... and it's for a different box, of course.

I found this innards pic of the Cleanbox 2, and it has a lot more resistors than the DTI drawing suggests... along with a couple of caps for a presumed Zobel across the output.

[Arpangel]

The ART Clean Box range are one of the most useful things you can have in a studio, I’ve got two CB Pro's, and I could immediately justify buying at least four more.
These things are like essentials in your studio tool box.

[ef37a]

https://artproaudio.com/framework/uploads/2018/06/om_dti.pdf

Ah... yes. Familiar with that one. It's a bit of an over-simplification I fear... and it's for a different box, of course.

I found this innards pic of the Cleanbox 2, and it has a lot more resistors than the DTI drawing suggests... along with a couple of caps for a presumed Zobel across the output.

You mean I woke up in me chair at 1:30 (a common occurrence these days!) and struggled for to find that info JUST to find you already had it?...!! Heh!

Yes, I could see it was a bit lacking but was at least a CIRCUIT diagram and not a block impression.

The two figures for insertion loss give a clue to the internals not being just transformers and also will only be correct for use the right signal direction.

Same applies of course to mains transformers but peeps should NOT play with those unless qualified.

I have a further question about audio transformers re loading etc Hugh but it will have to wait, slept in and must catch up!

Dave.

[Martin Walker]

https://artproaudio.com/framework/uploads/2018/06/om_dti.pdf

Martin, taken me a heck of a time to find that link again! Unfortunately no CR values but Self gives a 1:10 transformer mic pre into an op amp with a 100k load and 220pf in series with 22k.

You could start with that then check the response at 20kHz? I would not use a square wave, probably ring like billyo!

However, you are using the transformer as a DI box? What are you driving it with? If guitar as is usual then I doubt the response will still be flat compared to that found on the 'test rig'?

Thanks for that info Dave - I haven't really used this DI box much over the years, but feel more inclined to plug some line-level electronics into it at highish levels and then use the tranny to warm things up in a creative way.


Martin

[Martin Walker]

A quick check on its frequency response using the PluginDoctor utility (Dirac Pulse test signal) suggests that this rises significantly above 1kHz.

Yes, that's quite common.

My AP test system defaults to an input impedance of 200k Ohms, and that's so high as to be effectively an open-circuit load for a lot of transformer outputs. The result is almost always a high resonant peak around 50kHz or so.

Load the output with 10k Ohms and the response usually flattens out... and often loading with 600 Ohms introduces a gentle HF roll-off.

So when you come to experiment with your own taming Zobel network, make sure you also load the output to match whatever it normally plugs into, otherwise you might find you over-correct!

Ironically I'm already doing that, testing the DI box by inserting it between my Fireface UC phones output and one of the mic inputs, which is quoted as having an input impedance of 2k, so I was fully expecting the result to be suitably loaded.

Low-end response of the Fireface UC mic inputs @44.1kHz is quoted as a very useful -0.5dB @18Hz, while the phones output is an excellent -0.1dB at 1Hz and 20.4kHz, with an output impedance of just 30 ohms.

However, this is the response I've initially measured :shocked:

I suspect the software is having a minor fit testing hardware, as the top end of this curve doesn't shift sideways on screen if I alter the sampling rate as it does when testing software plug-ins.

Anyway, out with my boxes of resistors and tiny capacitors ;)


Martin

[ef37a]

My further questions Hugh when you have a mo.

There are two basic types of audio 'line' transformer. Those designed for output use and must be driven from as low a source impedance as possible. They have low winding resistances and are often bifilier wound.

Then there are input '10k-10k' bridging transformers which have much higher winding resistances and high inductance to as to present a high impedance at an input (correctly loaded of course). They are not suitable for output duty, often have an inter-winding screen and are the bees knees for solving serious RFI problem.

The transformers available from Art, Orchid and others SEEM to be of the output only variety but I do not recall reading any information about that?

Are they then some sort of compromise specification?

Dave.

[Hugh Robjohns]

I'm afraid I don't know what specific transformers are employed in the ART units (or other isolators).

I do know that Lundahl make a range of 'general purpose' transformers for use where it's not clear whether it's being used in an input or output configuration, and which they recommend for ground isolating applications!

https://www.lundahltransformers.com/general-purpose/

[ef37a]

Ah, so Lundahl are are at least straight enough to tell us that transformers cannot ultimately be both a high level output AND input device.

I can only surmise they are the sort of transformers that are sold for general isolator use?

Dave.

[Hugh Robjohns]

I have seen transformer units for which the instructions recommend placing as close as possible to the destination's input, rather than the source's output... and a few that recommended the other approach, too! It really all depends on their role. If it's unbal>bal then I'd generally place them at the source output, and if it's bal>unbal I'd usually go for the destination input.

If it's a ground loop issue I also generally place them at the source output... but some experimentation as to which sounds best is always a good idea!