4-channel gain stage with DC-servo for MiniDSP

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I'm painfully aware of the MiniDSP's output voltage limitations. I have some consumer amps that specify 1V RMS input voltage to reach full output power. More input voltage is needed to enter the "dynamic headroom" territory, too. Since we all know that digital clipping should be avoided, it seems that a prudent solution is an analog gain stage after the MiniDSP, so I designed one.

This uses the LME49740 (a quad version of the LME49720/LM4562) for the gain stage and the TL074 for an (optional) DC servo. The gain is set to 10dB, which should provide additional headroom and gain, and even allow some pro amps to be used. This could easily be increased to 20dB (10x gain) if desired.

I haven't properly set up the servo yet, but here is a preliminary schematic (attached), omitting things like bypass caps. This is one of four identical channels - use one board for each 2x4, two boards for each 2x8. This seems like something that many, many people are asking for. I can use some myself, so I plan to do a run of PCBs if there is enough interest.

Your comments and suggestions are of course welcome.

-Charlie
 

Attachments

  • MINIDSP GAIN STAGE WITH DC SERVO.JPG
    MINIDSP GAIN STAGE WITH DC SERVO.JPG
    43.1 KB · Views: 1,312
Account Closed
Joined 2001
Charlie,

Have you actually found the 0.9 VRMS output of the miniDSP units to be a problem/limitation?

I know, on paper, this would prevent some amplifiers from achieving their rated outputs, but most of us seldom (if ever) need to use that rated capability.

Cheers,

Dave.
 
  • Like
Reactions: 1 user
Charlie,

Have you actually found the 0.9 VRMS output of the miniDSP units to be a problem/limitation?

I know, on paper, this would prevent some amplifiers from achieving their rated outputs, but most of us seldom (if ever) need to use that rated capability.

Cheers,

Dave.

YES, definitely. Let me give you a specific example:

I recently built a 3-way open baffle system and used an Adcom GFA-2535 on each side. Because boost was being applied to the woofer at the low end of its operating range, I needed all of the amp's output capability. The woofer was an 8 ohm unit of only 85dB/W sensitivity, and I bridged two of the channels to provide 200W to the driver. Checking the manual, I found that in bridge mode the input sensitivity for the amp was 1.75V RMS (!) to reach the rated 200W power. With the MiniDSP driving the amp I could not really reach full power using the 4x10 (unbalanced). There was definitely no "extra" output voltage available for driving the amp a little harder, for transients, etc. Just to keep up with the other drivers, I had to use all of the dynamic range of the MiniDSP (operate all the way up to 0dB internal levels) on this channel, which didn't make me feel at ease either. I basically had to turn everything up to "max" but because I was using source material with some dynamic range to it, which caused the average SPL to be rather weak. I needed additional gain after the MiniDSP and before the amp(s).

So, yes - I think that there is definitely a need for a gain stage like this.

-Charlie
 
Account Closed
Joined 2001
Charlie,

You might consider adding an optional capacitor location across R2 which would allow to create a 6db/octave shelving filter....for possible dipole EQ (or maybe something else.) That allows to "unload" a possibly large EQ section out of the miniDSP and should simplify (possible) gain structure issues.

Cheers,

Dave.
 
Charlie,

You might consider adding an optional capacitor location across R2 which would allow to create a 6db/octave shelving filter....for possible dipole EQ (or maybe something else.) That allows to "unload" a possibly large EQ section out of the miniDSP and should simplify (possible) gain structure issues.

Cheers,

Dave.
Good point. I'll add that. If nothing else the cap can be used attenuate the gain at HF.

To do a shelving filter you need to add one more resistor in series with the cap. You can do a shelving filter very easily in the MiniDSP as well, and can change the shelving corners as needed, but I guess it would make just as much sense here where you have available gain. My Active Crossover Designer tools can calculate the biquad coefficients for a first order shelving filter if you want to implement one in the MiniDSP.

-Charlie
 
Here's a question - according to my sims there is a pretty good DC startup transient until the input DC blocking capacitor charges up. With the values shown, this seems to take a couple of hundred milliseconds before settling down to low levels. The peak magnitude is the product of the level of DC offset present in the input signal and the gain, but it is not something that you would want to reach your drivers.

One solution is a NC relay connected between the output of the gain stage and ground. After a few seconds delay, power is applied and the relay opens, removing itself from the circuit. This would require some additional circuitry and several relays, which seems expensive. Another option would be to use an LED-LDR (e.g. Vactrol) in series with the 1k resistor (R4) just after the DC blocking cap. Initially the LDR resistance is well over 1MOhm, and it's a simple matter to design a circuit that ramps up the current to the LED and slowly drops the resistance down to a couple of hundred ohms. Since the voltage drop across the LDR resistance element (and thus the current) is low in "normal operation" when the LDR resistance is low, its distortion will also be very low (see postscript), so under normal operation is should "disappear".

Does anyone happen to know the nominal level of DC offset output by the MiniDSP?

-Charlie

PS: here is a link to some distortion measurements for various LED-LDRs done by member jackinnj:
http://www.diyaudio.com/forums/analog-line-level/177332-comparison-light-dependent-resistors.html
 
Last edited:
Account Closed
Joined 2001
The miniDSP units already have DC blocking capacitor on their outputs so no (steady-state) DC will appear at your circuit input. You may have some DC on your circuit output, but it probably won't be much and could be reduced by using FET-input op-amps.

If the op-amp were in an inverting configuration you'd need another resistor in series with the capacitor. But you're showing a non-inverting configuration. Any first-order shelf can be achieved by using/adjusting just three components....R1, R2, and capacitor.

Cheers,

Dave.
 
If the op-amp were in an inverting configuration you'd need another resistor in series with the capacitor. But you're showing a non-inverting configuration. Any first-order shelf can be achieved by using/adjusting just three components....R1, R2, and capacitor.

Cheers,

Dave.

You are correct, but only if you want the shelf to transition between "full gain" and 0dB gain. If you want to transition between "full gain" and some other smaller gain, you need another resistor in series with the cap. The additional resistor makes the shelf implementation more flexible.

The miniDSP units already have DC blocking capacitor on their outputs so no (steady-state) DC will appear at your circuit input. You may have some DC on your circuit output, but it probably won't be much and could be reduced by using FET-input op-amps.
Hmmm... without the DC servo, my circuit sim shows the gain stage having only about 1.5mV of DC offset for 10dB gain. Most amps are AC coupled anyway, so this level of DC offset should not be a problem.

If the MiniDSP is AC coupled, maybe I should just pair this down to the bare minimum - no servo, no input coupling cap, and try to balance the input impedances to reduce the offset. I would definitely keep the input LP filter, however.

Actually, in that case, I already developed a board for the LME49270 that can implement all sorts of HP, LP filters, LT, etc. I should just use that! Some details can be found here, in a thread from 2011:
http://www.diyaudio.com/forums/anal...ilities-i-want-your-input-11.html#post2478905
What's ironic, is that I made a run of these boards and then put them aside after I discovered the MiniDSP, in which you can implement many circuits without having to actually "build" them. But now it seems my analog circuits will be useful afterall! I can implement a tailored "boost" circuit like LT or LF-boost shelf using the analog boards and do the other filtering in the MiniDSP for an analog-digital hybrid crossover. During development I can model the whole thing in the MiniDSP and then move only those circuit elements that require lots of boost into the analog domain.

-Charlie
 
Last edited:
Charlie,

Have you actually found the 0.9 VRMS output of the miniDSP units to be a problem/limitation?

I know, on paper, this would prevent some amplifiers from achieving their rated outputs, but most of us seldom (if ever) need to use that rated capability.

Cheers,

Dave.

Yeah, I just came around looking for a thread about this as I've just been working on a naked driver dipole and I keep having issues with gain.

The problem is audible distortion before amplifiers even get close to rated input. With 2 MiniDSP units driving the multichannel input on a receiver, and the inputs on an EP2000 for subs, I can't even get close to clipping before the MiniDSP starts making unpleasant noises.

They should just go ahead and make a version of these with some actual voltage capabilities so we don't have to build gain stages and ride them way past the point of linearity to drive an amplifier to full output.

MiniDSP's with actual discrete output stages would be great.

Charlie said:
.....MiniDSP and then move only those circuit elements that require lots of boost into the analog domain.

Nah, I'd rather just have a DSP that wasn't anemic, I don't really want to have to put 3 preamplifiers after the MiniDSP just to have the capability of flicking the clip indicators should I choose to...a feat any $300 receiver can do with it's subwoofer output, and so should these DSP boards.
 
Account Closed
Joined 2001
They should just go ahead and make a version of these with some actual voltage capabilities so we don't have to build gain stages and ride them way past the point of linearity to drive an amplifier to full output.

They already do. The 2x4 balanced version supports 2 VRMS (unbalanced) output voltages vice the "standard" unit which is limited to 0.9 VRMS. It sounds like you purchased the wrong device and/or your gain structure is still not correct.

Cheers,

Dave.
 
They already do. The 2x4 balanced version supports 2 VRMS (unbalanced) output voltages vice the "standard" unit which is limited to 0.9 VRMS. It sounds like you purchased the wrong device and/or your gain structure is still not correct.

Cheers,

Dave.

I guess my whole point is they shouldn't make a device at all that's incapable of driving 80% of amplifiers past a few watts of output while at the same time showing it being used in that configuration in the product literature and advertising :p I don't like riding gear to the point of non-linearity..so my choice is purchase LabGruppen FP14000 amplifiers so that I can get 100w out of them with half a volt of input, or only buy drivers that are greater than 95db efficient.

I don't mean to sound like a turd, but I've got computer sound cards that are ~$15 dollar el-cheapo and have three times as much output voltage. Gain structure isn't even an issue, unless it's considered normal to have to have a completely additional gain stage device like a preamplifier before AND after a MiniDSP board.

Perhaps I purchased the wrong units, but I'm still really asking...for whom would these units be the correct units in that case? I mean, these units start reaching audible levels of distortion at somewhere around 0.6 or 0.7v it seems, so I have to ask if anyone actually owns a power amplifier that can deliver it's rated power at that input level. Gain Structure Shmain Structure! LOL!

Sorry, I'm just really grumpy with these things. They could have put something other than one microscopic transistor in the output stage and provided a jumper that allowed a range of output voltages which could drive anything.
 
Account Closed
Joined 2001
Many amplifiers nowadays are spec'd with a voltage gain of 29db (x28). With a 0.9 VRMS limitation your amplifier should produce 25.3 VRMS into the load. That's approximately 80 watts into an 8 ohm load. (That's a considerable amount of power for most applications.)

You're incorrect that the miniDSP units produce audible levels of distortion at 0.6-0.7 volts. Only when you get to the clipping point at 0.9 volts does that occur.

You must have some other variable at work in your scheme that is causing a problem.

Cheers,

Dave.
 
Last edited:
They already do. The 2x4 balanced version supports 2 VRMS (unbalanced) output voltages vice the "standard" unit which is limited to 0.9 VRMS. It sounds like you purchased the wrong device and/or your gain structure is still not correct.

Cheers,

Dave.

But don't you have to run balanced input to the board and at 2Vrms in order to get the 2Vrms output?
 
No. You can run unbalanced input at either the 0.9 or 2.0 volt jumper setting and still get 2.0 volts output unbalanced.

miniDSP Balanced Kit | MiniDSP

Dave.

Thanks Dave and looking at the user guide (which I should have done first) I see how it all works.

The problem with the balanced board though is that it has op-amp buffers on the inputs (makes sense as it's balanced) and the outputs.

If I'm going to put the signal though additional stages I'd rather limit their number and also have control over what op-amps I used and how they were powered etc.

As such a buffer stage with gain and only on the output of the unbalanced board starts to makes sense.
 
Account Closed
Joined 2001
I believe miniDSP recognized the need for (possible) further voltage capability quite a while back and that prompted the balanced version. Yes, it does add extra circuity relative to the direct interface of the ADAU1701. For users who don't like the selection of particular op-amps or circuitry implementation it's certainly possible to add. However, you do lose the compact/integrated format of the OEM boards.

Different strokes for different folks I guess.

Cheers,

Dave.
 
That's pretty much exactly what prompted my almost aggravated comments, it doesn't seem like it would have been too hard to offer, say, 3v of output capability within the same form factor.

*shrug*

That's probably not possible with the current minimum power supply specification of 5VDC. That corresponds to +/-2.5V max supply voltages and the output devices need some "extra" voltage to operate correctly. This is probably why there is only 0.9VRms available (about 2.6Vpk-pk).

They could up the minimum power supply voltage to 12VDC, use a DC voltage inverter IC like the ADM660 to generate the negative rail, and then use a low noise analog gain stage to boost the output voltage swing capability to, say, +/-5Vrms (a typical max output level for analog preamplifiers).

This would require only minimal design input. Even I can design an analog gain stage... it's a mystery why they don't get to it, when many of their customers are begging for it and lack of output voltage is one of the main complaints about their platforms??? It's a big fail on their part IMHO.

-Charlie
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.