Mic Pres Based on THAT SSM INA chips

Various companies offer all in one mic pre solutions based on instrumentation amplifiers. Some examples include THAT1510, SSM2019, and INA217.

Curious if anyone has done any comparisons among them.  Interested in any objective or subjective impressions that wouldn't necessarily show up on the spec sheet.

And maybe the TI/Burr Brown PGA 2500 digitally controllable?

Please?

I built a reference design INA217 board several years back, and then after retrofitted it for a THAT chip.  Could not tell the difference, both sounded great.

Good to know regarding the 217 and that chips.

There's pga2500 looks too be 5V supply,  so headroom could be an issue.

There's pga2500 looks too be 5V supply,  so headroom could be an issue.

Click to expand...

It's +- 5 V supplies and balanced/differential, so not too bad

john12ax7 said:

Good to know regarding the 217 and that chips.

There's pga2500 looks too be 5V supply,  so headroom could be an issue.

Click to expand...

If that was an issue, we'd all be in trouble, as all ADDA's are either 5V powered or 3,3V powered...

The PGA2500 is digitally controlled, as in real analog gain control. That's why I'm interested. It's what RME uses for the FF400, but not in the FF800.

cyrano said:

If that was an issue, we'd all be in trouble, as all ADDA's are either 5V powered or 3,3V powered...

Click to expand...

Not the same thing as a mic front end though. While very high voltage rails in SS might not really be called for - except for marketing purposes - there is something to be said for good headroom and levels in the front end. For one thing many (most ?) well performing active components / circuits perform technically better with higher voltage rails.
By the time you hit the ADC it's all under control and you can set the level to suit.

I did a stint at the parent company of Toft consoles. The preamp section is based around the all in one chip set. We started off with the that 1510's and moved into the INA 217's.  The that's in the circuit had issues were they would die more often.  So we switched over to the INA217 and had less chip death in that circuit. I never worked out why but they were sonically similar.    Since it is compatible I did try the SSM2019 in there. There was a big  sonic different. The SSM 2019  had more clear top end but nearly as much in the bottom as the INA 217's  in turn the INA217's had a more smear in the top end but bigger bottom end.  Was unusual for sure.

I suspect that these chips don't have robust rail clamping on their dies (this is the first place to skimp when you are trying to shrink dies), which is why the reference designs call for extensive diode clamping on the inputs.  Typically one might consider these optional but on these chips they seem to be mandatory.

Do mean rail clamping for overvoltage protection?

Yup.  Many chips have extensive guard rings and ESD diodes on the sensitive inputs...others do not.

Matador said:

I suspect that these chips don't have robust rail clamping on their dies (this is the first place to skimp when you are trying to shrink dies), which is why the reference designs call for extensive diode clamping on the inputs.  Typically one might consider these optional but on these chips they seem to be mandatory.

Click to expand...

In mic preamps there is a known problem from the phantom blocking capacitors being shorted to ground when charged up to 48V. This throws a -48V spike on the preamp input. This transient can involve amps of current even from properly sized input caps. Of course the DIY belief that more must be better will sometimes oversize these phantom caps with bad result.

THAT corp published a white paper on the subject (Phantom menace IIRC) and may have even made a die change on their preamp ICs in response to that experience.
http://www.thatcorp.com/datashts/AES5335_48V_Phantom_Menace.pdf
http://www.thatcorp.com/datashts/AES7909_48V_Phantom_Menace_Returns.pdf

I recall having a conversations with a senior THAT corp engineer about a semi-custom version of one of their standard ICs involving just a metallization layer change, to rearrange the precision resistors to perform a different function. They declined to leave even a single I/O pin unprotected by on-chip clamps and whatever (my design change would have left one pin unprotected). 

JR

I've lost more than a few INA prototypes that were meant for dynamic microphone use (aka. no phantom power):  they just seem very susceptible to over/under voltage and ESD-like events.  It would appear that THAT is better at this than TI though.

Robust clamping requires a *lot* of die area.

Those new(ish) low capacitance bi-directional TVS diodes used for DSL lines might be good here. Instead of 4 diodes to rails, it would be 2 diodes to chassis ground.

Those new(ish) low capacitance bi-directional TVS diodes used for DSL lines might be good here. Instead of 4 diodes to rails, it would be 2 diodes to chassis ground.

Click to expand...

Just wondering about TVS capacitance. Looking at the SSM2019, fig 4 schematic, it already has 200pF across the input pins, so what would a low cap TVS offer, doesn't more capacitance offer better adsorbtion. (e.g.ONsemi CM1213 @ 0.85pF)

Martin Griffith said:

Just wondering about TVS capacitance. Looking at the SSM2019, fig 4 schematic, it already has 200pF across the input pins, so what would a low cap TVS offer, doesn't more capacitance offer better adsorbtion. (e.g.ONsemi CM1213 @ 0.85pF)

Click to expand...

The point of the diodes is simply to shunt currents like the "Phantom Menace" scenario JR cited. A small amount of capacitance is fine and will help absorb transients but that's not the point of them. Most TVS diodes are pretty high capacitance (some are several nF) which can result in HF loss and instability. The low cap ones have an extra signal diode in series which I guess isolates the larger zener junction?

I actually have some Bourns 24V CDSOD323-T24C-DSL and the 12V breakdown version CDSOD323-T12C-DSL which are ~3pF  but I have not actually tried them (datasheet says 3pF at 1MHz but nothing about audio frequencies). I would want to study their breakdown / distortion characteristics. TVS diodes are a relatively new concept so they have not been used for audio until recently. SMBJ15CA was used in at least the Rane SM26S but interestingly they are regular high capacitance ones. I guess Rane stuff tends to target industrial applications and they wanted to just make it indestructible. Even at 2.2n the source impedance would have to be >3k before it starts to dip into HF. But it is a little insensitive to expect upstream gear to drive 2.2n.

Would you test to IEC6100-4-2?  http://technodocbox.com/docs-images/74/70826092/images/12-1.jpg  I would have thought a simple passive LPF should take care of most of it. Are schotty diodes too leaky, but fast enough?

Martin Griffith said:

I would have thought a simple passive LPF should take care of most of it.

Click to expand...

A tiny cap can handle a quick transient spike yes. But that isn't going to handle the "Phantom Menance" scenario where the inside of those phantom power coupling caps is suddenly -48V causing 1A+ to be pulled from ... somewhere. To absorb that you need diodes.