New THAT 1570/5171 Digitally Programmable Mic Pre Design.

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
how lovely. a 2 chip solution.
Looks like a fair competitor, however, DIY'ers are going to have fun with QFN packages on this one.

http://en.wikipedia.org/wiki/QFN
 
Yes, QFN is not a very nice footprint to work with. However, turning the chip upside-down and solder small leads from it to the PCB makes it a bit easier. The chip needs to be glued to the PCB.

It sure seems to be a useful IC.
 
Rochey said:
DIY'ers are going to have fun with QFN packages on this one.

yes, after reading your thread about the "solder paste blues" i'm not enthused about working with SMD parts especially QFN packages.  

dusty circuit's solution does however bring encouragement.  

EDIT: upon some thought, Rochey's Digitally Controlled mic pre input PCB seems much simpler utilizing easy to obtain parts while providing the most necessary of utilities. 
THAT's 5171 would be interesting for other applications though.
 
 
dissonantstring said:
[EDIT: upon some thought, Rochey's Digitally Controlled mic pre input PCB seems much simpler utilizing easy to obtain parts while providing the most necessary of utilities. 
THAT's 5171 would be interesting for other applications though.
Not to thread steal - that board is designed to go in front of a That157x design, or a PGA2500.
Or... if allyou want is sexy switches on your panel, it'll go nicely with any mic pre.
 
JR,
I'm pretty sure that THAT already has curves for both parts. ;)
I didn't check it against a 2500, but it compares (very) nicely with the existing preamp chips out there.
The 2-chip implementation is a nice solution:
The 1570 (analog preamp) is built without any internal feedback resistors.
The 5171 (stepped resistor array) provides both the shunt AND feedback resistors for setting the gain.
The R-values of the stepped resistor array (5171) have been chosen to give the best noise performance throughout the gain range.
A built-in servo (ext caps needed), and a zero-crossing detector on the 5171 complete the deal.
The 1570 can be used stand-alone as a "regular" preamp. THAT has some adapter boards for those who don't want to deal with the small SMT pkg.
"Official" unveiling at AES (with demonstrations).

JP


 
someone build a module around the fet switch itself its goin to be bestseller, both in synth, and pro audio realms.
 
Looks pretty good.

I liked the PGA 2500 pretty well except for one thing...it's a +/- 5V chip.

This pre can run from 18v, so an extra 10db or so dynamic range.
EIN is about the same for both.

I wonder about the price. PGA 2500 is an expensive chip.

Les
L M Watts Technology
 
They posted budgetary pricing on their website.

The +/- 5V is not a big deal on TI part since majority are feeding 3-5V A/D convertors, but extra swing is icing on the cake, for digital controlled analog.

The zero cross detection is a nice feature.

JR
 
someone build a module around the fet switch itself its goin to be bestseller, both in synth, and pro audio realms.

This FET switch is very specifically designed for the 1570 preamp; the internal resistance values are selected for best noise/distortion at each gain step.
Plus, it is permanently configured as a dual-feedback/single-shunt array, so you're limited as to what you can do with it (not that people won't try  ;D )

JP
 
Perhaps a precision attenuator/pad

I like the values for bipolar mic pre...

and I could use this feedback network with my own discrete design...

while I suspect digital control will only appeal to specific product segments.

JR

 
JohnRoberts said:
They posted budgetary pricing on their website.
The +/- 5V is not a big deal on TI part since majority are feeding 3-5V A/D convertors, but extra swing is icing on the cake, for digital controlled analog.
The zero cross detection is a nice feature.
JR

±5V is always something that engineers bite on though. You can always add a small pad in front of the part, if your really convinced that your going to get huge signal swings like that.
Zero cross is par for the course, and I'm really not so sure about the price. Seems a little steep. PGA2500 lists for $7.95 these days.

For those of you that know me... yes I have some bias  ;D

But if I take the job hat off... and talk a little more like an audio designer (I hardly consider myself to be a *good* one)

I struggle to understand why people would go for a 2 chip solution, in a package thats difficult to prototype, at a cost which is higher than the incumbent.
The only advantage I see with this device, so far, is that it supports a higher input voltage rail, but with a simple pad, that's taken care of.

Did I miss something fundamental here?

/R
 
Rochey said:
I struggle to understand why people would go for a 2 chip solution, in a package thats difficult to prototype, at a cost which is higher than the incumbent.

I can see a few reasons, albeit nothing fundamental.

The combined solution (including passives) appears to need somewhat less PCB area than the 2500. The QFN is not much of a big deal even for small prototype labs these days.

Going for two chips presumably allows them to use a more-analog process for the front-end and a more-digital process for the control chip. It looks like this helps them to get a few dB better low-gain noise performance than the incumbent + a pad.

For me the #1 reason would be flexibility. A 2-chip solution gives me a bit more placement/routing flexibility, which can help in keeping the analog and digital domains separated on a dense board. Like John said it might be interesting to use their control chip with my own front-end, too.

Having said that I'm in no hurry to respin my existing designs from the 2500 to this chipset. It's always nice to have an alternative, though.

JDB.
[disclaimer: all this is backed by nothing more than back-of-the-envelope calculations based on a quick read of the relevant datasheets]
 
Rochey said:
JohnRoberts said:
They posted budgetary pricing on their website.
The +/- 5V is not a big deal on TI part since majority are feeding 3-5V A/D convertors, but extra swing is icing on the cake, for digital controlled analog.
The zero cross detection is a nice feature.
JR

±5V is always something that engineers bite on though. You can always add a small pad in front of the part, if your really convinced that your going to get huge signal swings like that.
Zero cross is par for the course, and I'm really not so sure about the price. Seems a little steep. PGA2500 lists for $7.95 these days.
Who knows what the real price is ? A serious customer will not be buying onsey twosey.
For those of you that know me... yes I have some bias  ;D

But if I take the job hat off... and talk a little more like an audio designer (I hardly consider myself to be a *good* one)

I struggle to understand why people would go for a 2 chip solution, in a package thats difficult to prototype, at a cost which is higher than the incumbent.
Difficulty to prototype is a mild irritation but generally ignored by serious designers.

Breaking out the resistor array makes it attractive to folks who want to provide the capability of digital control but differentiate their product with their better than the next guy's discrete preamp.  The analog preamp chip is a free standing product with a market of it's own.
The only advantage I see with this device, so far, is that it supports a higher input voltage rail, but with a simple pad, that's taken care of.

Did I miss something fundamental here?

/R

The beauty of digital control is that in concept you could provide smart control of input gain. Scaling and switching in a pad somewhat reduces the power of digital control. I suspect one of the output lines could control a pad, if it had a way to detect overload (next feature to add?).

The zero cross is useful... I could do the gain control with 3 sections of digital pot, but the zero cross is more glue than I would spend in a console so attractive.

The extra output swing while meaningless to the electrons, is not wasted on customers, who have opinions about such things.

JR
 
With the PGA2500, multiple chips are cascaded using the data in/data out pins to load all control bits. (only 4 pins from the MCU is needed to control no matter how many PGA2500 chips you have) . Though one can rig your schematic so each PGA is individually addressable, and not have to deal with the shift registers (data in/out).

The THAT uses individual Chip Select pin to address which chip needs to be controlled specifically. Each THAT chip requires a separate pin from the MCU to control the chip select. (6 pins from MCU needed to control 3 THAT chips, add another pin for each THAT chip).... though I guess one can use a multiplexer to reduce pin count.  Personally, I like this better. I only want to talk to the chip I want to address.

In terms of gain control, The PGA have the 7th and 6th MSB always 0.
So only bits 0..5 in sequential order is needed to control gain. (But of course, there's that big 10dB jump in gain on N=1.)  and you have that Gain(dB) = N+9 formula to deal with.  But you have 0dB gain capability though at N=0.

The THAT ignores bits 1..2.  (gain unchanged from N=1 to 7 ) There's a big 8dB jump, and the lowest gain is approx 6dB @ N=0 (Not 0db like the PGA2500).  Gain(dB) = N + 6 formula (approx).

I don't see any Overload bits on the THAT chip, unlike the PGA2500. (yeay! +1 to the PGA)
The OVR pin on the PGA2500 can be used to automatically decrement the N number to reduce gain by 1dB each time an overload condition is detected.

In terms of packages, I like the PGA2500 better.

 
Hmm, I think I'd prefer the chip that has an extra 10dB of headroom to the chip that turns itself down when it starts crapping out.

The two-chip setup opens this chip up to applications beyond what the 2500 can do. Some applications outside audio (data aquisition, imaging, sonar, etc ) have either no need of digital control or need different gain increments that would make a single-chip setup unusable.
 
The extra output swing while meaningless to the electrons, is not wasted on customers, who have opinions about such things.

Well, I'm running both PGA2500 and THAT preamps.
I gladly accept the extra 10db of dynamic range of the latter at high gains. When I pad it back i'm padding down the noise floor too...

But software control of gain and attenuation is mighty handy. I'll want to build up some stuff with the new chips. Looks like they would be pretty good as line recievers too if the external feedback resistors are configured for that.

Pretty expensive though, aren't they?

Les
L M Watts Technology
 
chip that turns itself down when it starts crapping out.

The chips don't turn itself down.

But you can write a program that detects if it's clipping and automatically decrement the gain by 1dB.

 
owel said:
The THAT uses individual Chip Select pin to address which chip needs to be controlled specifically. Each THAT chip requires a separate pin from the MCU to control the chip select. (6 pins from MCU needed to control 3 THAT chips, add another pin for each THAT chip).... though I guess one can use a multiplexer to reduce pin count.  Personally, I like this better. I only want to talk to the chip I want to address.

It looks like the THAT parts have a 3 bit programmable address so you should be able to address 8 of them with one chip select line.  Their data organization is less than clear but thats what happens when analog guys try to talk digital...

JR

PS: having a chip that automatically turns itself down is not always desirable, but detecting O/L could be useful in a system context where you can restore the gain in a later stage harmlessly. For recording there may be some merit to a persistent clip indicia that can be reset after a take, or whatever.
 
Back
Top