Old: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
Locked
ricardo
Posts: 262
Joined: Sun Apr 10, 2011 7:24 am

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by ricardo »

mediatechnology wrote: About the low gain offset: This circuit doesn't work very well: viewtopic.php?f=6&t=256&start=26

I don't know why yet, but you really need to offset the THAT1510/INA217 this way. I've tried both and this offset trim works no matter which servo you use:
http://www.ka-electronics.com/Images/jp ... ematic.jpg
The aforementioned circuit has lousy LF response at gain though. The input R and Cfb should be much, much larger.

With the "new and improved" improved servo, and low-gain Vos trimmed at the THAT1510/INA217 pin 5 connection, switching is relatively click-less.
There's no differential to common mode conversion of correction due to differential resistance.
T-bias can be used.
The servo correction range is doubled.
Wayne, does the first mentioned low gain offset circuit not work very well just for this Direct Coupled Preamp?

Or are you saying it doesn't work very well for An Improved Servo for the THAT1510 and THAT1512 too? ie the circuit with input capacitors.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

Welcome ricardo.
There have been at least 3 different ECM8000 circuits.
I can believe that.
I had been told or read that the ECM8000 pulls all of its current out of one leg.
That does not appear to be the case here but there is a significant (and correctable) current imbalance.
Wayne, does the first mentioned low gain offset circuit not work very well just for this Direct Coupled Preamp?

Or are you saying it doesn't work very well for An Improved Servo for the THAT1510 and THAT1512 too? ie the circuit with input capacitors.
I found this one: http://www.waynekirkwood.com/images/jpg ... ervo_1.JPG difficult to adjust properly no matter which topology is used - not just a capacitorless circuit.
I suspect that by offsetting the integrator and taking it back into the input subjects that term to variable preamp gain.
Offsetting the ref pin, pin 5, appears to keep the output offset constant regardless of gain.

One thing I did try - but didn't use because it has a slight CMRR penalty - is to connect a 10R from pin 5 to ground and inject current into the 10R.
That saves an op amp.
You only need +/-5 mV range to correct the 1510 output offset.
The added 10R hurts CMR a little bit with a 1510. (In some cases it might improve it.)

EDIT: Thought it might be instructive to go back and revisit why the ref pin needs a trim. From: viewtopic.php?f=6&t=256
What if the THAT1510 or 1512 were used to measure it's own offset?

Remember that the worst-case output offset is defined, in millivolts, as (0.25*Gain)+/-5 mV. The (worst case) 250 uV of Rgain terminal voltage amplified by the 1510 gain allows us to improve, at 60 dB gain, our DC measurement capability 1000-fold. 250 uV at the Rgain pins becomes 250 mV at the output.

and...

At low gains however the 1510s internal output amplifier offset, the "+/- 5 mV" term, works against us. Although clicking at high gain is virtually eliminated, clicking at low gain is quite severe. The servo "sees" offset at Rgain which does not really exist. The key is to make the 1510 DC-accurate at low gains.
What we are trying to do is null the 1510's differential output amplifer's offset, +/-5mV, independently of the servo.
By (1) defeating the servo (2) setting the 1510 to unity gain, (3) reading the 1510 output offset, and (4) setting the ref pin trim for 0V (<<100 uV) output the 1510's output offset term is greatly reduced. (The input Vos contribution is small.)

Though it would seem to be the same, offsetting the integrator doesn't accomplish the same result because the correction is fed back into the input, not Vref.
It's no longer independent of the servo, the +/-5mV correction becomes part of the servo.
That 5 mV of correction applied to the integrator then gets multiplied by the gain of the preamp.
I've never found a way to not make it click by eliminating the ref trim.
If the minimum gain is allowed to remain relatively high, maybe +20 dB, it's less of an issue.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Preliminary Common Mode Servo Scehmatic

Post by mediatechnology »

This is the Input Capacitor-less DC-coupled Mic Preamp's Common Mode Servo and Flying Rail Supply. I'm using this one for testing.

Image
Direct Coupled Input Capacitor-less Mic Preamp Common Mode Servo and Flying Rail Generator

Notes:

The input has a passive pole with a 3.4 Hz corner frequency; I may need to tune this. This is for a relatively fast response with a tau of ~ 50 ms.

Op Amp "A" is connected as a voltage follower.
The MJE243 and MJE253 bootstrap the supply voltages.
I'm looking at some higher-voltage TO-126 parts (these are 100V) to provide a little more margin. The pickings are somewhat slim: MJE15034/35 and 2SC3503/2SA1381 are candidates.
The 16V Zeners and 10K resistors set the flying rail voltages at ~ Vcm +/-15V to provide 30V supplies for the common mode servo itself, the THAT1510 preamp IC and the differential servo an OPA2277.
The positive flying rail, with 0 to +48V common mode voltage, ranges from ~ +15 to +63V. The negative flying rail goes from -15V to +33V.
This keeps the THAT1510's power supplies and it's input common mode range centered about the microphone's quiescent "Q-point" voltage.

The base-collector capacitors, 100 nF/100V, provide stabilization.
Various 100 nF capacitors provide bypass.

My current op amp choice is an LME49720 dual to provide a low output impedance vs. frequency.
Though bipolar, the LME49720 has a significantly lower bias current than a 5532. The 1 nA bias current does develop an error across the 100K but it is not significant.
I've also used 5532s and the rather large ~1V error from Ib*Rin does not have significant impact on circuit operation.
My concern was that DC-precise op amps, such as the OPA2277, or BIFETS with low bias current, might not have low output impedance across that audio band.

No build-out resistance is required since the op amp is not driving any significant capacitance.

An optional differential bypass capacitor can be used.
I have not experimented with using electrolytics in that location.
Both ends of that bypass float together - but 30V apart - and essentially bootstrap it.

I've raised the Zeners in the past to permit the 1510 to run on 36V but the LME49720 has a 34V maximum.
I will be trying the 44V part, the LME49860.

Total current draw, not including phantom, is around 22 mA for the entire floating section comprising the THAT1510, LME49720 and OPA2277.
This works out to be about 2W most of it appearing as heat in the MJE243/253 and 10K resistors.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by JR. »

If I were to put this inside a large console I would want a more efficient power supply for heat and energy management.

Warning this is just a concept and no solder has been melted to prove that this works. but it seems like it should work to generate a nominal +/- unregulated supply that can float to any DC voltage that the floating supply is anchored too and components can tolerate (caps and diodes).

So a real voltage source will need to anchor these and accommodate any current in these rails that don't net out to zero, but that phantom supply already exists so this should be relatively straightforward.
floater.gif
floater.gif (4.78 KiB) Viewed 14333 times
Just to splain the theory of operation, this is a variant on a cap doubler, but rather than charging a string of boost caps to ground referenced reservoirs, the two boost caps get reset by each other to the nominal Vp-p of the transformer. The negative rail boost cap gets an extra FW rectifier to deal with the different polarity involved between negative and positive absolute output voltage.

Of course a non-polar cap could simplify that somewhat. If the current draw is low even a film cap could be used... but for production diodes and polar caps are cheap.

JR

PS: If nobody is interested I guess I could build one to prove it out, but i don't have a preamp to connect it to... I have some high voltage caps laying around that I ordered years ago, when I was more active with this project.
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

If I were to put this inside a large console I would want a more efficient power supply for heat and energy management.
Definitely.
I used a series of stacked 24V supplies for now to keep focus on the preamp.

I've got a little Murata 3W switcher on the bench right now that's giving me +72V and +24V from 12V input.

Murata NMT1272SC: http://www.ka-electronics.com/images/pd ... verter.pdf

It's a little noisy without an output pi filter.
With 2.5mH and 1 uF I can get it under 40 mV p-p.

I think there's a lot of ways to skin that cat.

I'll sort out what you sent.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by JR. »

Digital switchers have been showing up with more frequency (pun intended) in consoles, while Ok for a healthy sized 48+18V rail in combination with individual floating pass regulators. I'd be disinclined to add two switchers per mic input. FWIW there is a class of power supplies that use boost caps similar to mine, but switch those caps at HF so smaller caps can be used. I rolled my own that way once to make a phantom supply using a couple of small .1uF caps.

Since these floating rails are not regulating but just shifting a nominal DC voltage up to some different floating potential, a common HF clock could be used so there would be no beating or interference between channels. If this rail is not drawing much current per input, the 60 Hz approach is reasonable.

Note: I might be tempted to add two diodes in series with my positive boost cap so there is a similar number of drops in both directions, while the supply needs to be anchored and balanced actively anyhow.

========
Not to throw in another veer, it seems the output of such a floating preamp could be DC coupled too. While a simple differential amp couldn't handle the DC voltage (I've scratched out a discrete 3 input differential design), feeding resistors from preamp + and - output to a pair of simple inverting opamps, then sending another resistor from the floating 0V to a third inverter. Then resistors subtract that inverted version of the flying 0V from the two audio stems creating a DC 0V referenced output that is DC coupled.

Note: I do not suggest that this is worth doing, only that it "could" be done. Also any DC current drawn from my floating boost would need to be accounted for.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by JR. »

I have been amusing myself recently thinking about this (still), it is a decades old pursuit, while not very hard pursuit. :roll:

I have often thought of hanging an A/D convertor off the flying preamp so it can perform the conversion wherever it is.

Typical IC A/D convertors are Single supply with DC input offsets (roughly 2V for a decent TI a/d).

The TI programmable mic pre (2500) allows for biasing the preamp relative to that 2V A/D DC bias. While they still proudly cap couple their inputs, since that now becomes 2V +/- a diode drop.

I really like the idea of digital gain control for an ultimate preamp. TI, actually suggests degraded S/N and THD+N from the DC connection, actually it's the N that now is not padded down. They suggest pad and cap coupling. :shock: In fact I am temped to add a cheap micro for digital servo, and automatic clipping protection, etc.

The THAT 5171 could work biased up to 2v but the internal servo and probably zero cross detectors would not be happy. :(

I need to think about this some more. I have been leaning toward rolling my own front end around some best of class discrete transistors (new improved version of 737) with modern uber opamps. With PNP input devices self biased through the feedback resistors, I could pretty easily hit nominal 2V DC operating point.

There is no reason why a DC coupled pad could not be added if it really delivers a measurable "N" improvement.

This is more mental masturbation for now since I do not have a market for my DC coupled A/D convertor but i remain enthusiastic.
I don't want to reinvent the low distortion switched resistor arrays for gain setting, while they are clearly optimized for 30v p-p. and in the case of the 5171 0V dc bias.

JR
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

There's no reason the PGA2500 or THAT1571/5171 (or 5173) couldn't be flown.
I've thought about how the SPI would be opto-coupled.
The "Universal Input" would not require a pad.

I'm not sure flying the A/D is all that worthwhile for a couple of reasons:

1) I know it's getting to be "all-DSP-all-the-time" (I am seeing it used more often than not as a tool to muddy analog-centric posts) but a significant majority of end-users still want to hook up analog outboard gear right after the preamp.
This would require A/D-D/A double-conversion and add latency.
"Yes, this preamp has an A/D. No, there's no analog insert."
I can assure you this would not "fly" with an end-user.
Most people have their own favorite converters anyway and it would eliminate that choice as well.

2) Eliminating that film cap on the output may appeal to the fringe but most people's first requirement in a mic pre (or after one) would be an analog HPF.
Having a HPF in the digital domain provides benefit only after the unwanted LF energy has eaten up analog and digital headroom.

I think just because something can be done doesn't necessarily mean it should.
I'm staying on-course and letting people make their own post-preamp processing and converter choices.

I'm going to work through your power supply.
There's not a "one-size-fits-all" solution but it looks like a good idea to pursue.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by JR. »

I should preface this as mostly a mental exercise on my part, as has been my inclination for years and years.

I have posted in the past that I would like to also roll my own A/D design. I am slowly coming to the realization that I could not improve upon something like the PCM4222 without a brain transplant.

If we discuss how much of this is just proving that something could be done, vs actually having demonstrable merit if would be useful to put numbers to both. That should nicely demonstrate some measurable benefit from small servo caps vs, large electrolytics, etc. Note: Sam Groner has done some high resolution measurements of THD in modern electrolytic caps in different configurations, while I don't recall any biased up in phantom DC blocking configuration.

My gut says yes they (electrolytic) introduce measurable errors (with a good enough bench), but we all know that typical phantom powered mics are not pure and without farads (and virtually unmeasurable with high enough resolution).

If this blocking cap was very audible, we could pretty cheaply DC couple preamps when phantom supply is turned off, with a few more switch poles. (this is unlikely to happen because of consumer sensitivity to scratchy gain pots and switch clicks)
========

Sorry I am not trying to be argumentative (its just my nature)..I can't control where my subliminal design team takes me. :lol: Basic audio path technology is pretty mature. I have long been inclined to think this had more legs as a marketing exercise, than a significant advancement to the SOTA, thus my investigation of expanding the capacitor removal.

JR

PS: A flying insert could be provided with transformers. Surely the OTB pukes who insist on analog efx would not complain about quality transformers in that path, only for the inserts at that. While transformers kind of moot the whole flying supply theme... :lol:

PPS: another old idea that people don't much care for, is driving pin 1 negative with a servo that maintains DC coupled phantom inputs at 0V DC. 48 V Phantom supply would be the total voltage with pin 1 - V reversed out from modest positive phantom supply. The nice thing about this is that the preamp remains ground based. I expect that flying pin 1 (negative) would cause far more problem interactions with existing products than other proposed approaches (hot mic bodies, etc). Perhaps tolerable for an esoteric product, not mainstream compatible.
Cancel the "cancel culture", do not support mob hatred.
User avatar
mediatechnology
Posts: 5442
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: A Direct-Coupled Input-Capacitorless Active Mic Preamp

Post by mediatechnology »

I should preface this as mostly a mental exercise on my part, as has been my inclination for years and years.
I'm way past that stage with this.
If we discuss how much of this is just proving that something could be done, vs actually having demonstrable merit if would be useful to put numbers to both. That should nicely demonstrate some measurable benefit from small servo caps vs, large electrolytics, etc. Note: Sam Groner has done some high resolution measurements of THD in modern electrolytic caps in different configurations, while I don't recall any biased up in phantom DC blocking configuration.

Well, there was this back on page 11: viewtopic.php?f=6&t=14&start=108
I have long been inclined to think this had more legs as a marketing exercise
I don't.
than a significant advancement to the SOTA
A dismissive tone filled the room...

I don't really view this as a THD-reducer though it might be because it does eliminate three electrolytics two of which have DC-bias and fairly large stored charge. The third electrolytic is quite large in value.

I think the real benefit may be - because I've actually built one rather than imagined it - slightly lower noise from lower LF source impedance and leakage current as well as the elimination of DA and stored charge effects.

Talk is cheap.
Locked