PGA2500

[jdbakker]

Is there any way I can talk you out of using the MSP430 ?

I've spent some time getting reacquainted with the MSP430, and while it's a very nifty architecture, it has a few drawbacks for this particular application. First, the PGA2500 is a 5V device but the I/O cells on the MSP430-F2013 aren't 5V tolerant, so a direct interface is out. Furthermore, the IAR Kickstart IDE / C compiler that comes with the eZ430 is a limited version. For DIY usage it would be nice if a free-of-charge full compiler were available, but none appear to exist for the MSP430 at the moment (there is a GCC port underway, but it appears rather alpha-quality ATM).

May I suggest the ATtiny26 ? Its feature set is comparable to the MSP430-F2013 (sans the 16-bit SD ADC, but with more I/O pins). It will run directly on 5V and it has a mature free-of-charge GCC port avaiilable.

Do let me know if any of my facts are off. I'll admit to having more hours on the AVR than on the MSP430, and I can see the attraction in using TI parts, but for this application and a DIY context I'm not sure it's the best tool for the job.

[quote author="Rochey"]BTW - don't you have a PGA2500 EVM?[/quote]
Sure do.

JDB.

 

[Rochey]

hmmmm... TMS470 ... ARM7

Sure, use whatever - however, can I suggest that you make the code portable to any processor with a decent C compiler if possible.

It'd be nice for those of us who aren't very processor driven to understand what the code is doing as well.



/R

 

[Rochey]

by the way, worth mentioning that the "limited compiler" is open up to 4K in program size. Seeing as most of the devices have less than 4K of flash, it shouldn't really be a limit

the 3.3V to 5V is though, however, as one chip would be driving multiple 2500's I would have put a driver in there anyway.

/R

 

[jdbakker]

[quote author="Rochey"]Sure, use whatever - however, can I suggest that you make the code portable to any processor with a decent C compiler if possible.[/quote]
Right. I was planning to abstract the interfaces away, so the majority of the code would be directly reusable on different processors. All you'd need to do is change the low level definitions ('this I/O bit is the clock line, this I/O bit is the data line' etc).

On a similar vein, you may want to have the PGA and the CPU on separate boards. That's better from a noise POV, and it allows you to expand the system to pretty much any desired number of channels with just one CPU.

[quote author="Rochey"]by the way, worth mentioning that the "limited compiler" is open up to 4K in program size.[/quote]
Yeah, I saw the 4k figure, but 4k of what ? Code ? Preprocessed code ? Linker output ? Stripped linker output ? Neither Google nor an admittedly superficial scan of the docs yielded a definite answer. I have no direct experience with IAR, but tool manufacturers have been known to play fast and loose with such specs. All I know is that GCC has no such limitation, and microcontroller firmware tends to grow to fill all available memory ;-)

[quote author="Rochey"]the 3.3V to 5V is though, however, as one chip would be driving multiple 2500's I would have put a driver in there anyway.[/quote]
It's not the lines from the CPU to the PGA that I worry about, but rather the PGA's digital output lines (most notably the overload line).

JDB.

 

[jdbakker]

[quote author="I"][quote author="Rochey"]the 3.3V to 5V is though, however, as one chip would be driving multiple 2500's I would have put a driver in there anyway.[/quote]
It's not the lines from the CPU to the PGA that I worry about, but rather the PGA's digital output lines (most notably the overload line).[/quote]
...and you'd want access to the overload line to implement a dynamic gain scaling algorithm like PRR has described in this thread.

JDB.

 

[Rochey]

2 things....

you can't treat the PGA2500 like VCA, swining the attenuation up and down that quickly. However, lowering the amount of gain gradually when you begin to clip can be done with some logic.

Please ensure you can switch off that scaling. Some applications require the gain to remain fixed.

/R

 

[jdbakker]

[quote author="Rochey"]you can't treat the PGA2500 like VCA, swining the attenuation up and down that quickly. However, lowering the amount of gain gradually when you begin to clip can be done with some logic.[/quote]
The original thread was about relay switching, which isn't as fast as VCA swinging either. I'm well aware of the PGA2500 zero-crossing detector and its longish timeout.

[quote author="Rochey"]Please ensure you can switch off that scaling. Some applications require the gain to remain fixed.[/quote]
Hey, I'm most likely not going to implement it; I prefer fixed gain with lots of headroom. I'm not saying you should have scaling, just that you could have it.

JDB.

 

[Rochey]

have you seen the RME Micstacy?

they have it done pretty wisely, set the gain at the start of the gig, and if a channel gets to loud, it'll lower the volume, but never increase it.



/r

 

[SSLtech]

VW car radios do the same thing. If you set it too loud it'll turn down and stay there... doesn't stop you turning it back up in the quiet bits, but woe-betide you if you crank it when it's loud. -It just turns it down and says "that's silly, and I have the last word".

Excellent. Personally I hate listening to clipping, so that's fine with me.

Keith

 

[Rochey]

Keef,

You've volunteered for the power now... you might regret this

can we spec a power supply with enough grunt to supply:

// 8 PGA2500's
...... ±5V - (40mA each)Analog 320mA per supply.
...... -5V Digital (2mA)-- Total 16mA

// 32 relays (12V or 24V Digital) - (12mA each@12V = 384mA

// +5V digital -- unsure of the power. Used for logic, CPU, LED's and switches.

// 48V Phantom

If any device need +3.3 then a local LDO can be used from the 5VDD.

And All of this from a 24VAC Wall wart? (the kind that JLM uses?)

 

[chrissugar]

Just to put more gas on the fire, I should mention that the preamp based on PGA2500 I mentioned on page 2 in a direct shootout with Cranesong Flamingo, the Millennia HV3 series pre, and the DAV pre, sounded cleaner, richer and closer to the acoustic performance.
Now you can understand why I'm so interested in finding a simple way to control the gain of this preamp. :grin:

Come on Rockey. Do something. Pleeeezz. :grin:

chrissugar

 

[lagoausente]

[quote author="mcs"]I had a look at the datasheet. Programming the PGA2500 is very easy, and not a problem for any DIY'er with a bit of logic/programming experience. The small control panel I made for my relay pot circuit could easily be used for this purpose, if anybody wan't to try the PGA chip...

You can also get an evaluation board which includes programming software - but how much does that cost?

But I probably won't be testing the PGA2500 anytime soon. I don't really have a use for it - I like having a big rotary switch for gain selection...

Best regards,

Mikkel C. Simonsen[/quote],


Finally, is there any way to build a preamp based on this chip for a DIYer without programming knowledge?
What about the evaluation board?

 

[Rochey]

Hi,

The control interface for the PGA2500 is over SPI. SPI is the simplest serial protocol there is... you have data, a clock that is at the same rate as the data and a "word clock".

There is no way to control the PGA2500 without SPI.

SPI can be generated from almost any processor these days.

The MSP430 (www.MSP430.com) has example code to generate SPI data for you. however, the difficult task is actually controlling the data in the first place.

I'm sorry, I don't have a quick and easy solution for you at this point.

Thanks again

Rochey

P.S. the Eval board uses the PC parallel port to control the mic pre.

 

[rolo95]

Hey guys
im new in here...

love this thread....
can a Microchip PIC Microcontroller be used for the SPI communication ?

 

[Rochey]

yes - although I prefer the msp430 from an official point of view (I work for TI), and, even with my TI hat off, from a programming point of view.

I got the $20 dev tool from TI (_and_ I paid for it myself) and was up and running with the blinking LED (the equivelent of the hello world!) within about 15 minutes of opening the box.

I like it when stuff is easy

 

[rolo95]

Txs rochey

However programing is not my bread and butter

But i have a friend who are very good at it
let me see if he can help us with the code...

So the TI chip will be, then can you post links of all of the project documentation so far and what is still pending to do...

Greets.
Rolo.

 

[Svart]

I'm working on building up protocol state machines in VHDL..

Maybe I should give this a try..

 

[e.oelberg]

Hi

I have a prismsound orpheus, it uses pga2500, same like in apogee emsemble and rme fireface 400. It sound very very good. Better for most stuff than my outboard pres and miles better than metric halo 2882 pres.
It worth !

nicholas

 

[rolo95]

[quote author="Svart"]I'm working on building up protocol state machines in VHDL..

Maybe I should give this a try..[/quote]

Yeah

My friend... is swamped on a project ... until mid of 2009 :?

Hope you can code easily on the TI MC....

ask someone if they had an assembler for it

 

[jensenmann]

We´re currently testing a digital multicore system at work which uses these chips as preamps. Gain can be controlled from the headamp gain controller of our Yamaha consoles through the network while preamps (8x+ADs) remain on stage. It´s a pretty cool system with very high redundancy.
more infos (sorry no schematics or source code) here: medianumerics.com

Soundwise the designer wanted wire with gain and I think he reached what he wanted.
Since out installation is brand new and a bit complex (live concert hall + studio + OB truck outputs) I can´t comment about how it sounds in conjunction with our studio setup. We´re able to send every signal from the hall to the studio and backwards, so thats what´s going to be tested in November while we´re doing the first bigger live-shows.