Edirol UA-5 mic pre upgrade problems (phantom power)

[ben9000]

Hi all, Ben from Australia here, this is my first post here. Please excuse the length of it but I feel like writing about this conundrum might help me as well as you guys understand what's going on. Okay - I've got an Edirol UA-5 dual pre/audio interface which I use in stand-alone mode for field recording. It's main shortcomings are the quality of the pre's (not enough gain, last 25dB of gain bunched up between 4-5 o'clock on the pots, not 'robust' sounding). So among other more minor tweaks, I've made up on perfboard some INA163-based pre's. I followed the datasheet example schematic and they function fine as individual modules, with lots of gain, very low noise, and a nice transparent sound. I've made up a 12-position switched gain control which is lovely and even and lets me match the channels perfectly. So that's all good. The problem arises integrating this double-pre board into the UA-5. I've mounted the board inside the case and powered it etc. I can inject the output of the new pre via a 47uf electro cap into the AD chip (ak4524) with no problems. Clean as a whistle. I've removed the original surface mount blocking caps (their printed labels read line by line "1","50S" and "3KN" - can anyone shed light on this?) thus disconecting the stock input circuits from the XLR's. I've taken the p48 source (after the stock 6.8k resistors) and sent that up the XLR's on pins 2+3. This is blocked by 2 22uf 63v electro's (I have also used 47uf 50v non polar electros). All this is fine - it basically works perfectly with self-powerd and dynamic mics. But as soon as phantom power is engaged all sorts of strange phenomena begin. It acts as an AM radio reciever. It acts as a theremin, with several nasty cross-modulating tones rising/falling depending on the proximity of my hand. It will magically stop (or reduce, depending on it's mood) these symptoms if I touch the case of the two 9V batteries that power the pre's. They also disappear if I place a wire link between the metal outside shell of the microphone end of the XLR cables. On the other hand, sometimes this will have no effect at all, or the opposite will be true. Meanwhile, the pre's are still basically functioning, but will cut out when gain is bumped up to about 15dB away from max. When the gain is backed off, they will come to life again. When phantom is turned off and settles down (takes 30 seconds - rather long?) it all returns to normal. I have given up exploring all of the permutations of this behaviour and am now just trying to figure out why p48 has this effect. Do any of these things sound like they have an immediately apparent cause? I don't know where to start to resolve this problem. I will continue to check that phantom is being blocked from the pre inputs properly, but if any of you could shed some light on this problem it would be greatly appreciated. I'd be happy to provide you with as much info about the UA-5 as I can to help the sleuthing.
Cheers, Ben

 

[Kev]

welcome

difficult to follow a long string of words like that ... A diagram of your circuit layout could be good ... or picture.

Seems that the phantom is leaking to the input of the chip ... and sending it to a rail.

These problems aren't there on the work bench with the new INA163-based pre's but manifest only when inside the Edirol UA-5 and connected to the AD chip (ak4524) ?
:?
The AM radio thing must be telling us something.

does your retro PCB have a 0 volts connection to the main PCB ?

 

[burdij]

While a schematic of the front end would be helpful, I will ask some general questions.

Are all the supplies bypassed with good quality electolytics and monolythics? Are they bypassed to the common analog ground?

Do you include high frequency compensation in the input op amp feedback path? You could limit the frequency response to <100KHz without degrading the audio in any way, since your max sample rate is probably 96Khz or less.

 

[Ralf Biermann]

I think the problem might arise with electrolytic caps, since they have a high leakage current and cause a problem with the bias point of the pre-amp.

Just use polypropylene caps.

Best Regards
Ralf

http://musicology777.com/index.htm

 

[Kit]

"Words are flowing out like endless rain into a paper cup..."

Welcome to the forum, Ben.

I would like to help you but you lost me halfway trough the post......maybe a schematic could clear things up?

 

[gyraf]

(repost, spaces added - Jakob E.)

Hi all, Ben from Australia here, this is my first post here.

Please excuse the length of it but I feel like writing about this conundrum might help me as well as you guys understand what's going on.

Okay -

I've got an Edirol UA-5 dual pre/audio interface which I use in stand-alone mode for field recording.

It's main shortcomings are the quality of the pre's (not enough gain, last 25dB of gain bunched up between 4-5 o'clock on the pots, not 'robust' sounding).

So among other more minor tweaks, I've made up on perfboard some INA163-based pre's. I followed the datasheet example schematic and they function fine as individual modules, with lots of gain, very low noise, and a nice transparent sound.

I've made up a 12-position switched gain control which is lovely and even and lets me match the channels perfectly.

So that's all good.

The problem arises integrating this double-pre board into the UA-5:

I've mounted the board inside the case and powered it etc.

I can inject the output of the new pre via a 47uf electro cap into the AD chip (ak4524) with no problems. Clean as a whistle.

I've removed the original surface mount blocking caps (their printed labels read line by line "1","50S" and "3KN" - can anyone shed light on this?) thus disconecting the stock input circuits from the XLR's.

I've taken the p48 source (after the stock 6.8k resistors) and sent that up the XLR's on pins 2+3. This is blocked by 2 22uf 63v electro's (I have also used 47uf 50v non polar electros). All this is fine - it basically works perfectly with self-powerd and dynamic mics.

But as soon as phantom power is engaged all sorts of strange phenomena begin. It acts as an AM radio reciever. It acts as a theremin, with several nasty cross-modulating tones rising/falling depending on the proximity of my hand.

It will magically stop (or reduce, depending on it's mood) these symptoms if I touch the case of the two 9V batteries that power the pre's. They also disappear if I place a wire link between the metal outside shell of the microphone end of the XLR cables.

On the other hand, sometimes this will have no effect at all, or the opposite will be true. Meanwhile, the pre's are still basically functioning, but will cut out when gain is bumped up to about 15dB away from max.

When the gain is backed off, they will come to life again.

When phantom is turned off and settles down (takes 30 seconds - rather long?) it all returns to normal.

I have given up exploring all of the permutations of this behaviour and am now just trying to figure out why p48 has this effect.

Do any of these things sound like they have an immediately apparent cause?

I don't know where to start to resolve this problem.

I will continue to check that phantom is being blocked from the pre inputs properly, but if any of you could shed some light on this problem it would be greatly appreciated.

I'd be happy to provide you with as much info about the UA-5 as I can to help the sleuthing.

Cheers, Ben

 

[Ralf Biermann]

Hello,

I assume the configuration of the circuit looks similar to this:

http://musicology777.com/phantom.htm

The problem is the decoupling cap. When it is an electrolytic cap, it has a leakage current, which results in a parallel resistance to C1. So a dc current supplied by the phantom source flows into the input of the amp, which shifts the bias point.

There is no problem when the phantom source is disconnected with dynamical microphones.

Greetings from Germany
Ralf


http://musicology777.com[/url]

 

[Kev]

thanks Ralf

oh and
Welcome to the forum :thumb:

I forgot earlier ... sorry

Ralf's web page
he also is playing with harmonics
http://musicology777.com/

OH and and !!
thanks Jakob
... that made the point better than I did .. :thumb: :thumb:

 

[Ralf Biermann]

Hello,

there might be a second problem: a condensator microphone has a high impedance, therefore it is more sensitive to radio waves. The solution is to add a 100pF ceramic cap in parallel to the input, from the microphone to ground.

Greetings from Germany
Ralf

 

[ben9000]

Hi Kev, Ralf, Jakob, Burdij and Kit - thanks for your replies. Hopefully this is only temporary, but the whole project is off line at the moment because while I was probing for a suitable point to tap off 5v for my recorder the UA-5 shut down, and won't switch back on

And I'd been so careful up until now! Grr.

Getting it going again will be a whole new and "challenging" project, but assuming it leaps to life again at some point we'll forge ahead as before!

All of a sudden the Tascam HD-P2 is looking pretty good...
http://www.tascam.com/Products/hdp2.html

So anyway, back to it:

Jakob - The spaces helped thanks!

Kev - Yes, the new pre board has it's 0v point connected to ground on the UA-5. 48v is grounded to the negative pin of the 12v power socket. This appears to have different characteristics to the ground point mentioned above, and these connection methods give the best results.
Even worse inteference happens if they are swapped or connected to the same point.

Ralf - Yes, the phantom power part of the circuit is basically as in your diagram, where R1 is 6.8k and C1 is 47uf. This same thing is applied to both the hot+cold mic output signals. After C1 the circuit is as found here in fig 5:
http://focus.ti.com/lit/ds/symlink/ina163.pdf

I will look into polypropylene caps, but are these rather large? Space is tight in this application. Thanks for the tip about high impedance, I will try the small capacitor trick - but the mics I'm using are max 600 ohm, and the lowest is 150 ohm, so not too high.

Burdij - Re HF compensation - from my limited understanding of opamp pre design I don't think this is possible with the INA163 as it is not an opamp per se - it is an 'all in one' chip with a bunch of stuff under the hood. Maybe you are aware of this and I'm off the mark - in which case I'd like to hear more about this idea. There's definitely something odd happening supersonically!

Kit - The best I can do for a schematic is the link above and then point you to http://he.fi/photo/ua5-dissection/
Scroll down the page and select pictres 11-20 in 'huge' view. Here you will find in picture 14 the point that I am getting 48v from.

Just right of the combo XLR/TRS sockets you can see the 8 stock blocking caps, (4 of which keep phantom off of the TRS terminals).

These I have removed and the XLR's have been totally disconnected. Right of the caps in a parallel row are 4 x 6.8k resistors interspersed with transistors.

The left side of these resistors is my 48v source. Unfortunately I cannot determine how the UA-5 ultimately derives 48v. Does anyone see components in these photos which would be doing this? I've done some tracing but can't pinpoint anything.

I think that's all, so thank you all for your input, I'll keep you posted!
Ben

 

[sodderboy]

Welcome Ben!
Is pin 1 of the XLR's connected to the Edirol ground plane and to the ground of your circuit? Sounds like things are floating when P48 is engaged. Test to find 48 volts between pin1 and pins 3 and 2.
Mike

 

[ben9000]

Hi Mike,
Pin 1 is connected to the negative pin of the power supply socket, and not to ground of the new circuit board. Making this connection resulted in a 'throbbing' of the signal. The audio would drop in and out at about a 1Hz cycle. More gremlins!

However, ground on the new board IS connected to what appears to be signal ground on the Edirol. This gives a different result to connecting to the negative terminal.

I think I'll make my signature "A little knowledge is dangerous".

Ben