Balanced input +4dBu for bridged LM3886/LM4780

[phofman]

Hi, I have been trying to figure this out for the whole day to no avail, mostly searching this website...

I am looking for a stereo amp module(s) with about low-distortion 2 x 50W to connect to ESI Juli@ soundcard. The soundcard offers balanced +4dBu outputs, IMO DC coupled. The card will be hooked to a thin-client PC with only +5V in the PCI slot - I will have to make dual +12/-12V for the card. The PC, card, and amp will be in one case (separated, the PCI slot will be extended to a different section of the case). The PC has its own adapter with no earth wire, i.e. no ground loops to the amp.

IMO it would make sense to use the balanced outputs (when already provided by the card) and choose an amp with dual supply requirement - I will need -12V anyways.

I like digital amps (my previous project https://github.com/pavhofman/plabs-player/wiki used Tripath TK2050 ) but this time I am thinking of the flat low distortion chart of chip amps.

There must be a way to make single LM3886 with balanced input - https://www.jlmaudio.com/forum/viewtopic.php?t=21 . However, I have not been able to find any other example of such solution. I would not mind using a bridge (2 x LM3886 or 1 x LM4780 for one channel), yet I did not find any recommended circuit for the +4dBu level either.

I know I can use the differential receiver to return to single-ended, but then I can use single-ended outputs of the soundcard directly. I would really like to keep balanced down to the amp or to the speakers.

I have looked at the digital amps, perhaps a TDA7498 module could be modifed for balanced input +4dBu. However the datasheet distortion values are higher than those of gainclones and I would have to make the negative PSU anyway for the soundcard (although only low-power one).

My idea of PSU is two 24V 150W SMPS modules New AC 100-240V to DC 24V 9A Power Supply Board 150W AC-DC Switch Power Module | eBay , placed into a separate metal enclosure, properly grounded to PE.

Thanks a lot for your suggestions/recommendations/experience.

 

[tomchr]

You can always add a THAT1200 or go with one of my THAT Receiver boards.

Tom

 

[phofman]

Thanks, but as I said, I would like to keep balanced all the way. If not, I can always use the single-ended outputs of the soundcard.

Perhaps a chip amp is not good solution for this case.

 

[AndrewT]

Keeping a balanced throughput all the way to the speaker is admirable. The speaker is balanced and would suit that.
But what does that bring you?

There is nothing wrong with using the balanced facility for the interconnections and using unbalanced in the processing. If a balanced output is needed to send to the next stage then you can convert from unbal to bal for that connection.
You end up with:
Bal > unbal > process > unbal > bal
The big advantage of a balanced impedance connection is the extra attenuation of interference.
Is there an excess of interference that will affect your interconnection?

 

[phofman]

Hi Andrew,

I understand your point. Actually my plan to go balanced is more of a desire than technical need. I am thinking if the soundcard already offers good high-level balanced output, it would be a waste of resources not to use it.

I ordered a cheap dual LM3886 board (looks decent to me for first sight) HiFi LM3886TF Stereo Amplifier AMP Assembled Board 68W+68W 4? 50W*2 / 38W*2 8? and will do some measurements (with another Juli@), how it runs at balanced +4dBu with +/-24V PSU. IMO I will have to lower the gain to use full input range without clipping. I will post the results.

 

[AndrewT]

Hi Andrew,

I understand your point. Actually my plan to go balanced is more of a desire than technical need. I am thinking if the soundcard already offers good high-level balanced output, it would be a waste of resources not to use it.

Then it's OK to use it. Balanced impedance connections are used because they offer improved interference attenuation.
Balanced signalling inside an amplifier does not offer this.

I ordered a cheap dual LM3886 board (looks decent to me for first sight) HiFi LM3886TF Stereo Amplifier AMP Assembled Board 68W+68W 4? 50W*2 / 38W*2 8? and will do some measurements (with another Juli@), how it runs at balanced +4dBu with +/-24V PSU. IMO I will have to lower the gain to use full input range without clipping. I will post the results.

Using a high level input will drive the chip amp into clipping.
The minimum gain for the 3886 chipamp is stated as +20dB, but the stability margins for audio duty, at +20dB gain, are compromised.
Most Builders adopt >= +26dB and many recommend somewhere in the range +28dB to +30dB for chip gain.
If you adopt these higher gains for improved stability margins, then you cannot use a high level input.

There are a few ways to adapt a gain restricted opamp (and power opamp) to work well at lower gains. I used the noise gain version. I think Tomchr did some thing similar but he keeps much as a commercial secret, even though he is in a DIY Forum. I used +20dB gain in my only implementation of this noise gain trick for a desk top active monitor. Again I think Tomchr chose a similar gain value.

But a 60W into 8ohms chipamp starts to clip @ ~21.9Vac (for sinewave). 68W into 4ohms clips @ 16.5Vac.
A gain of +20dB requires the input to never exceed 2.19Vac (8ohms), or 1.65Vac (4ohms) and a +28dB version requires the input to never exceed 872mVac (8ohms) or 657mVac (4ohms).

 

[phofman]

Actually the peak level of that input is +20dBu (7.5V). I would need only 3x amplification (about 10dB), not a job for a chip amp...

I looked at OPA541 which is supposed to accept lower gain, not very common.

Would you perhaps know of any well-tested good quality discrete-components circuit for handling such low-gain power amplification? Like a FET power stage or similar...

 

[AndrewT]

Some of the Pass amps have fairly low gain.
A Power Follower has +0dB gain (same input and output voltages).

Very low gain amplifiers do have more of that common mode distortion. The solution to that is to adopt inverting instead of non-inverting.

Have a look at using the inverting (opamp) to attenuate the big input signal down to a more managable level.
But maybe the easiest solution is simply to attenuate.

 

[phofman]

I think I will use one half of the large balanced signal for lower-noise volume control and then attenuate for single-ended higher-gain power amp. I will try to keep the amp gain as low as possible (while keeping it stable). Either chip amp or some proven solid state amp, not decided yet. Thanks a lot for your help.

 

[sgrossklass]

I think I will use one half of the large balanced signal for lower-noise volume control and then attenuate for single-ended higher-gain power amp.

Ugh. That's a total bodge, the first part I mean. Not recommended. Whatever you do, always use a proper balanced input stage. Either the usual opamp job for an unbalanced circuit or a balanced opamp (e.g. OPA1632) if you are going balanced internally. Choose its gain for the desired internal levels.

I basically know what you mean with lower-noise volume control, but isn't the noise from a typical NatSemi chipamp easily low enough as-is? As in, stick-your-ear-to-the-tweeter kind of levels? Plus, higher level input signal also generally means higher-value pot, and we all know that impedance unbalance is not particularly beneficial.

 

[phofman]

Thanks a lot. From what I see the easiest and technically still correct will be to use the regular unbalanced consumer-level outputs of the soundcard, especially if the connection will be just a few centimeters long. Most likely relay volume attenuator and the signal directly to the power stage, no attenuation needed here. Basically the most common way.