THAT1570/5171 DIP Adapter

[mediatechnology]

Alex N just sent me one of his THAT1570 THAT5171 version 3 adapters for test. Thanks Alex!



The photo does not show the high build quality and finish he has gotten.

You'll notice two jumpers, J1 and J2, that he added for me. These break the Rgain lines between the 1570 and 5171 to allow the combo to perform magic tricks. If I'd had one of these earlier I wouldn't have to have done this:



I hope to have some time to test this little baby very, very soon. Then, we'll talk about the magic tricks.

[JR.]

Kool... I wish there more more of these dip adaptor available for many of the hip new chips that are only available in SMT... I miss the good old days when we could actually prototype circuits without having to layout a dedicated PCB.

Some of the new parts I like, are also liked by cellphone makers, so they only come in tiny ass packages.

JR

[mediatechnology]

I'm glad Alex did this. THAT have them for many of their SMT-only parts but the 1570 and 5171 adapters they have are separate. Alex's implementation is a little more practical. I hope he does well with these.

Have you looked at the Brown Dog adapters? http://cimarrontechnology.com/index.aspx

[JR.]

Yup, I've heard of them, but I don't see an adapter that would have reduced my pain with that class D audio amp chip, about the size of a hungry wood tick (with a heat sink pad on the bottom). I love the part but small and a PIA to mess with myself...

FWIW I just relaid my tuner PCB to use a shorter height package for my microprocessor, because the parts sit between the PCB and faceplate, so height matters. That new part was also about 1/3 the footprint size. It was a PIA to route because the small square pin-out can't be routed underneath, between the rows of pins, lots more vias.

JR

[promixe]

Hi All,

Here are some more pictures of the adapters I made, one for THAT1570/5171 combo, and the other one is for PGA2500:



The pinout of the THATDIP adapter - green is for 1570 connections, blue is for 5171, corresponding individual pin numbers as per each chip's datasheets are in ():



The PGADIP pinout is identical to PGA2500's datasheet pinout.

I am in the process of creating a channel PCB to house the THATDIP adapter and here is what I have in mind: Download Schematic PDF
The unbalanced meter output would theoretically go into a uC ADC for processing, hence the -20dB gain. This part is still work in progress... =)
The balanced output is ~+6.4dB to make total gain of the channel 20-75dB.

As I was telling Wayne, I've been concentrating on two layouts of the channel PCB concurrently: one is "U-shaped", meaning that I/O XLRs are next to each other facing back panel, and generally routing is going around the PCB in U-shape. This one is pretty small and self-contained, i.e. you assemble the board and no wiring needed beside hooking power and SPI to it. The second type I'm working on is a straight back-to-front routing, with choice of Combo Neutrik or plain XLR female input, getting to a 3-pin socket on the other end of the board to be routed into anything off-board. This board is longer, but narrower, so one could stack 12-16 of these into 1RU case. Also, I've already finished a complementary output PCB for the latter design, which has RFI caps/ferrites and a vertical stacking jack (NSJ12-type) to save horizontal rack space.

Also, I've been working on another version of the channel board that has a GPIO switchable insert headers in between the 1570 and the output driver, to accommodate optional insert modules, such as a compressor, or EQ, right on top of the channel PCB, and digitally controlled along with the 5171 of each channel. I'm yet to perfect the idea, but the PCB for it is 50% finished already. =)

I'm anxious to hear everyone's suggestions/improvements/ideas on any of the above. Thank you all in advance!

Alex

[mediatechnology]

Alex - Thank you for posting here and also for the better photos and pinout.

I'm struck by the odd similarity on grain pattern of our two desks. It's uncanny.

[JR.]

Hi All,


I am in the process of creating a channel PCB to house the THATDIP adapter and here is what I have in mind: Download Schematic PDF

Nice work.

From a quick glance you need to check the balanced output for at least one missing connection. I would also be inclined to reference the output driver to chassis ground also. You are sending a differential feed to the output driver so why not reference the output to ground at the output jack?

The unbalanced meter output would theoretically go into a uC ADC for processing, hence the -20dB gain. This part is still work in progress... =)

In my experience, this doesn't need to be buffered with an expensive opamp, while you want it to be a low impedance (if micro A/D input is sampled by S/H as most are). In general you can manage source impedance with a C to ground, while you don't want s/h noise to get back upstream into clean audio.

The balanced output is ~+6.4dB to make total gain of the channel 20-75dB.

As I was telling Wayne, I've been concentrating on two layouts of the channel PCB concurrently: one is "U-shaped", meaning that I/O XLRs are next to each other facing back panel, and generally routing is going around the PCB in U-shape. This one is pretty small and self-contained, i.e. you assemble the board and no wiring needed beside hooking power and SPI to it. The second type I'm working on is a straight back-to-front routing, with choice of Combo Neutrik or plain XLR female input, getting to a 3-pin socket on the other end of the board to be routed into anything off-board. This board is longer, but narrower, so one could stack 12-16 of these into 1RU case. Also, I've already finished a complementary output PCB for the latter design, which has RFI caps/ferrites and a vertical stacking jack (NSJ12-type) to save horizontal rack space.

Interesting, I would be inclined to make modules of x8 channels (perhaps x4), I am not completely sure what the utility of the DIP adapter is for actually making more than onsey twosey. The DIP uses roughly the same real estate as placing the part down on the PCB.

with x4 and x8 modules you could plop one micro processor per module to handle meters, and perhaps manage multiple SPI loops (generate individual chip select lines from the common micro, and accept input from some external source or common control surface)

Also, I've been working on another version of the channel board that has a GPIO switchable insert headers in between the 1570 and the output driver, to accommodate optional insert modules, such as a compressor, or EQ, right on top of the channel PCB, and digitally controlled along with the 5171 of each channel. I'm yet to perfect the idea, but the PCB for it is 50% finished already. =)

I'm anxious to hear everyone's suggestions/improvements/ideas on any of the above. Thank you all in advance!

Alex

Interesting... good luck...

JR

[mediatechnology]

Alex - Good work!

One thing I noticed - and it's probably an sch error - is that the output of U2B, pin 7, should connect to the RH side of R19 (at the junction of R19 and R22.) The output of U2B is hanging.

[promixe]

From a quick glance you need to check the balanced output for at least one missing connection.

One thing I noticed - and it's probably an sch error - is that the output of U2B, pin 7, should connect to the RH side of R19 (at the junction of R19 and R22.) The output of U2B is hanging.

Thank you Wayne and JR for pointing this out, I think you're referring to the same thing. Fixed!

I would also be inclined to reference the output driver to chassis ground also. You are sending a differential feed to the output driver so why not reference the output to ground at the output jack?

The additional output board I mentioned in my previous post looks like this:


If I understood you correctly, the output board accomplishes what you're talking about. Are you also saying to reference pin 1 of H2 to CHA instead of AGND (on the mainboard schematic PDF that is)?

In my experience, this doesn't need to be buffered with an expensive opamp, while you want it to be a low impedance (if micro A/D input is sampled by S/H as most are). In general you can manage source impedance with a C to ground, while you don't want s/h noise to get back upstream into clean audio.

The general idea behind this was to give additional flexibility to the end-user to be able to output a high quality unbalanced signal in addition to balanced. If the unbalanced audio output is not needed it could be either used for metering purposes or left unpopulated on the PCB. I think the opamp-less way you're mentioning is really handy and saves some PCB real-estate (and $), but how well would it work as a main unbalanced output of the channel vs. having it opamp-buffered? And also, to be honest, I'm not 100% clear as far as how to implement your suggestion correctly. =) I need to study more on that one...

Interesting, I would be inclined to make modules of x8 channels (perhaps x4), I am not completely sure what the utility of the DIP adapter is for actually making more than onsey twosey. The DIP uses roughly the same real estate as placing the part down on the PCB.

You are right, the main purpose for the DIP adapter is to do lab work, prototype, or make low-quanitity DIY preamps. Ideally I'd like to perfect design of a channel board that I could fit into a 1RU with up to 16 channels in it. The main reason for it is mobility - I know quite a few people who would LOVE to have 12-16 high quality preamps in as small chassis as possible to use on their remote recording gigs. And if I design for 16 channels in mind, then fitting 4x channels is a piece of cake.

Thanks for your insights!

Alex

[JR.]

I would also be inclined to reference the output driver to chassis ground also. You are sending a differential feed to the output driver so why not reference the output to ground at the output jack?

The additional output board I mentioned in my previous post looks like this:


If I understood you correctly, the output board accomplishes what you're talking about. Are you also saying to reference pin 1 of H2 to CHA instead of AGND (on the mainboard schematic PDF that is)?

Pin 1 H2 stays chassis ground, or specifically chassis ground(2).
On the output driver chip, the schematic for that section is incomplete, but output reference ground R23, and pin 1 H2 connect together and to chassis ground(2). They do not need to connect to earth ground on the PCB more than once (assuming a metal chassis). You may want to add a 100ohm R between agnd and this local output earth ground so the circuit will still work when not mounted in a chassis.

In my experience, this doesn't need to be buffered with an expensive opamp, while you want it to be a low impedance (if micro A/D input is sampled by S/H as most are). In general you can manage source impedance with a C to ground, while you don't want s/h noise to get back upstream into clean audio.

The general idea behind this was to give additional flexibility to the end-user to be able to output a high quality unbalanced signal in addition to balanced. If the unbalanced audio output is not needed it could be either used for metering purposes or left unpopulated on the PCB. I think the opamp-less way you're mentioning is really handy and saves some PCB real-estate (and $), but how well would it work as a main unbalanced output of the channel vs. having it opamp-buffered? And also, to be honest, I'm not 100% clear as far as how to implement your suggestion correctly. =) I need to study more on that one...

To input the line level signal into a low voltage micro, you need to divide it down to a lower Vp-p (say 3.3V) and bias it up at 1/2 the micro's PS (1.6V). performing this pad and bias circduit with 3 resistors, just add a cap to ground at the common point forming a 20kHz or so LPF. This was adequate to deal with S/H current spikes at the A/D input in my experience, but this could vary with different processors.

Note: similar to my ground observation about the balanced output, this optional output differential (if used) needs to reference to the distant, not local ground. i,e, ground end of R27 and C24 connect to H9 pin 1. Again we probably want a resistor (10-100 ohm) between this distant ground reference and the local AGND. If we visualize what that stage is doing, it is dividing the signal down to 1/10th the input voltage, but keeping it referenced to the input ground, and relying upon brute force hard connection to make the external ground the same as the local ground. By forward referencing the differential amp to the external ground, it ignores voltage differences between the grounds, a good thing especially when dealing with a padded down signal. It also reduces corruption of the local AGND by external ground currents.

Thanks for your insights!

Alex

You are welcome...

JR