DEQ2496 consumer level analog board

stef1777

Member
2004-06-18 11:53 am
Paris
Hi Folk!

I started to design a small "consumer level" analog board for the Behringer DEQ2496. I need some advice to go further on this project.

The base idea is to build a small pcb to be connected directly on the HE10 connector on the DEQ2496 DSP board. A second HE10 connector permit to connect the cable going to the original input/output board. Most of the original signals pass true the pcb except the input and output analog signals for main left and right.

[IMGDEAD]http://www.awdiy.com/uploads/images/DEQ-Analog/carte_analog.jpg[/IMGDEAD]

I plan to use DRV134 and INA134 chips from Texas or one of their equivalent from THAT.

Beta schematics

My questions are the following:

- I'm loose with signal level and not sure what is the right chip to use for input and output (0dB, +3dB or 6dB version). Note that on the DEQ DSP board, the ADC connections are not direct and this circuit can not be removed. I inserted the circuit in my schematic.

- Do I really need to have a LFE circuit at AK4393 DAC output? I've read in the datasheet that the AK4393 already have an internal LFE circuit.

- Are the coupling capacitors well placed in my current beta design?

One important point: I want to stay with a minimum solution with the minimum possible quantity of components.

If you want to help, thank a lot.

Stef...
 
Hi Stef,
I'm not saying you're wrong, but the recommendations I have seen is for the decoupling caps on the op amps were always straight across the +15 to -15, without the gnd. connections.

Also, I think the caps on the analog inputs (47uF) wouldn't be needed unless there was excessive DC offset from the DRV134. Perhaps a 1 or 2 uF poly or bipolar would be better only if it's needed?

otherwise, I'd like to see a bit of gain in the output stage - I know the DCX2496 direct output voltage is a bit lower than I like.

just my thoughts. looks like a good idea though.
 
Pulse-R said:
just my thoughts. looks like a good idea though.

Some of your comments are already in beta3 schematic. Output is now +6dB. The PSU is far from the THAT chips, the 100nF/10uF couple is better right to the THAT pins.

DEQ-miniAnalog beta3

I still need to solve the lowpass filter problem. Still not sure for the input gain.

The board will be vertical if possible. Better for heat.

Stef...
 
another suggestion, can you run the inputs as more balanced, such as with + in to THAT +in (J1-1 to IC1.4, and J1-2 to IC1.3), and and say 10pf from IC1.3 to gnd? not sure if that's good, but would allow some lower noise/lower offset perhaps by lifting the input gnd. to cable gnd?

with the decoupling, what I thought was THAT cap pin 6 to 5 with 10uF low ESR? not sure if you knew what I meant.

input gain of standard analog I think is -12dB, same as DCX2496 input.
 
Stephan,

The ADC needs the signal centered on 2.5VDC. So you need to offset the outputs of the 1646 to be at 2.5V and indeed you don't need (don't WANT) the output coupling caps.

That is why in my DCX output board I used the OPA1632 for the analog input channels. This chip is specifically designed to drive balanced ADC's and has a pin for offsetting the output DC level.

You put 2.5V (from a simple reference) on this pin and you are done. I think you can do it with the non-inverting input pin of the 1646 but depending on the gain the reference will need to be lower than 2.5V of course.

Jan Didden
 

stef1777

Member
2004-06-18 11:53 am
Paris
HI Jan!

Thanks for your help.

I was not aware of the OPA1632. Pretty nice chip. The difficulty with it seems that I don't have a direct access to AINL+/- (resistor deviser present on the dsp board), Vcom (2.5v) is not present on the X1 connector too, its an SMD chip and it require many more external components. But the chip is really great and also have an Enable pin.

Currently, I don't know how to use it in an easy way with the DEQ but if you have some idea, I'm not sticked on the THAT1646...

Stef...



janneman said:
Stephan,

The ADC needs the signal centered on 2.5VDC. So you need to offset the outputs of the 1646 to be at 2.5V and indeed you don't need (don't WANT) the output coupling caps.

That is why in my DCX output board I used the OPA1632 for the analog input channels. This chip is specifically designed to drive balanced ADC's and has a pin for offsetting the output DC level.

You put 2.5V (from a simple reference) on this pin and you are done. I think you can do it with the non-inverting input pin of the 1646 but depending on the gain the reference will need to be lower than 2.5V of course.

Jan Didden
 
stef1777 said:
HI Jan!

Thanks for your help.

I was not aware of the OPA1632. Pretty nice chip. The difficulty with it seems that I don't have a direct access to AINL+/- (resistor deviser present on the dsp board), Vcom (2.5v) is not present on the X1 connector too, its an SMD chip and it require many more external components. But the chip is really great and also have an Enable pin.

Currently, I don't know how to use it in an easy way with the DEQ but if you have some idea, I'm not sticked on the THAT1646...

Stef...





Ahh! Yes, but the resistive divider on the input of the ADC does NOT divide the 2.5V. The 2.5V is present on both outputs so does stay the same. Seems strange, but observe: you have three resistors in series. At each end, there is 2.5V. So, at the junctions, you can not have anything other than 2.5V also. It is only the analog signal that is divided but that can be taken care of with some extra gain in the 1632. You would need that gain with an other solution anyway.

I use the 1632 in my DCX mod, and for the 2.5V I use an LM285Z-G fed from the pos supply. It is a precision 2.5V ref.

I can just manage to solder the 1632; any smaller things like Frank Oettle's digital DCX mod I have to pass :xeye: .
Indeed, you would need a couple of resistors externally, but it's a choice to make.

Jan Didden
 

stef1777

Member
2004-06-18 11:53 am
Paris
SOIC-8 is not a problem to solder. Space is counted on the board (something as 30x80mm). In fact, it's better.

Ok, if I understand, the resistive divider on the DSP board is not a problem. I just need to adjust the gain on the 1632.

I also have +5v (digital) on the pin 40 of the J2 connector. I can may be use it to produce the 2.5V?

I also have something as an analog mute signal on the pin 28 of of the J2 connector. May be it can be use to fit 1632's Enable pins? Is this interesting to ovoid "ploc" at startup?

On Figure 1 of the 1632 datasheet, I can see two 1nF (C1 and C2). Do I really need them?

I will design a new circuit later using the 1632 to see.


janneman said:


Ahh! Yes, but the resistive divider on the input of the ADC does NOT divide the 2.5V. The 2.5V is present on both outputs so does stay the same. Seems strange, but observe: you have three resistors in series. At each end, there is 2.5V. So, at the junctions, you can not have anything other than 2.5V also. It is only the analog signal that is divided but that can be taken care of with some extra gain in the 1632. You would need that gain with an other solution anyway.

I use the 1632 in my DCX mod, and for the 2.5V I use an LM285Z-G fed from the pos supply. It is a precision 2.5V ref.

I can just manage to solder the 1632; any smaller things like Frank Oettle's digital DCX mod I have to pass :xeye: .
Indeed, you would need a couple of resistors externally, but it's a choice to make.

Jan Didden
 
The caps are used to isolate cap loads but in the DEQ you have large enough series resistors of 470 ohms IIRC, so you don't need them.

Yes you can make the 2.5V from the 5V without any problems; it iss not critical. Even like 2.7 or 2.3 will work OK but your max output level is max with 2.5V. Or you can use a LED if you select one for close to 2.5V.

But if you use the 5V make sure you decouple the 2.5V to the 1632 with an R-C to get rid of any supply hum & noise.

Jan Didden
 
janneman said:
Yes you can make the 2.5V from the 5V without any problems; it iss not critical. Even like 2.7 or 2.3 will work OK but your max output level is max with 2.5V. Or you can use a LED if you select one for close to 2.5V.

Will use +15v to produce the 2.5v. The +5v available on the connector is the digital's one.

Do I really need the 1K resistors at input? Still trying to reduce parts... I may be have to add a caps at input to protect all this stuff.

I've read in the 1632 datasheet that the 1632 input is very sensitive to load inpedance. It need low inpedance input. Can be a problem with tube preamp?

They recommand 0.25 gain value in the datasheet? Seems low?

.
 
stef1777 said:


Will use +15v to produce the 2.5v. The +5v available on the connector is the digital's one.

Do I really need the 1K resistors at input? Still trying to reduce parts... I may be have to add a caps at input to protect all this stuff.

I've read in the 1632 datasheet that the 1632 input is very sensitive to load inpedance. It need low inpedance input. Can be a problem with tube preamp?

They recommand 0.25 gain value in the datasheet? Seems low?

.


Yes, I also use the analog supply.

You need the input resistors of course; you must set the gain by R5/R1 (and R6/R2). The input resistor should be below 2k for best performance. No problem normally but yes if you have a preamp with high Zout (or if you have a passive level control/preamp) you would need a buffer.

You should set the gain so that with a nominal input level your output can drive the ADC(with the attenuator in place, so you need at the output something like 3*2.3V pk, or about 5V RMS.
For a 2V nominal input, you can use the 1k for R1,2 and 2.5k for R5,6.

Jan Didden
 
janneman said:

You should set the gain so that with a nominal input level your output can drive the ADC(with the attenuator in place, so you need at the output something like 3*2.3V pk, or about 5V RMS.
For a 2V nominal input, you can use the 1k for R1,2 and 2.5k for R5,6.

Jan Didden

Thanks for advice.

The 1632 is nice but so many parts to add. :D

12 resistors + the parts to produce the 2.5v

.
 
What about the THS4131?

Seems to be the same but half the price and easiest to source.

I don't think that I'll have the space to place input and output circuits on the small pcb. As my first need is to have a "consumer" analog input board, I will may be only build input circuit and place a connector on the board to later add a high end output circuit if needed on another pcb. Seems more versatile to me. I've also found a way to add the board in the DEQ and use the current XLR connectors and relays (bypass) with only 2 components to unsolder.

Stef...
 
My last schematics.

DEQ minAnalog beta4

- removed output circuit. Replaced with an addon connector to later implement a separated output circuit.
- added Vocm circuit
- some changes on the input circuit
- added complet AK5393 input circuit

I'll try to add an input buffer (with LM4562) if I have enough space on the pcb. This is not a priority, I don't have high impedance pre-amp.

.
 
Hi Stephane,

Looks good. A few comments: the series R for the ref seems high: 330k. Are you sure there will be enough current to make it work well? Also, you may want to put a small resistor (100-1k) from the ref cathode to the decoupling cap to improve noise rejection.

When I look at the amps it seems that the 'gain' actually is 1/4. Is that what is intended? You will need a LOT of signal this way, also remember that the resistors on the ADC also attenuate by 1/3 the signal. Probably you wanted to swap the 270 ohms and the 2.5k resistors?

Jan Didden
 
Hi Jan!

Two very good questions from you.

janneman said:
Looks good. A few comments: the series R for the ref seems high: 330k. Are you sure there will be enough current to make it work well? Also, you may want to put a small resistor (100-1k) from the ref cathode to the decoupling cap to improve noise rejection.

In fact, I don't know how many current will use the two Vocm. I've found nothing on this in the 1632 datasheet. What have you done on your own design for the DCX?

In can try to use 100K and 1K and measure when I'll will have built a prototype.


janneman said:
Hi Stephane,

When I look at the amps it seems that the 'gain' actually is 1/4. Is that what is intended? You will need a LOT of signal this way, also remember that the resistors on the ADC also attenuate by 1/3 the signal. Probably you wanted to swap the 270 ohms and the 2.5k resistors?

Again a nice question.

We need to have the ADC/DAC run at maximum level for best performance. Input gain is for this really important. I'm not sure if the 1632 can manage the needed gain. May be a first buffer is needed.

BUT, the DEQ also have it's own internal RTA level meter screen with db or dbu print. If we enter, for example, a -40db signal, the screen must show -40db...

In the 1632 datasheet, they also write "The feedback resistor values (R3 and R4) should be kept relatively low, as indicated, for best noise performance."

As result of all these parameters to "manage", I really don't know how to select the right gain. Again, may be the prototype will help to answer all this. Or a guru of dbu. I'm so bad with math. ;)

Stef...


Ps : I made a mistake in the beta4 (AINR- and AIN+ are inverted). beta5 coming soon. ;)

Ps : what about THS4131?
 
stef1777 said:
Hi Jan!

Two very good questions from you.



In fact, I don't know how many current will use the two Vocm. I've found nothing on this in the 1632 datasheet. What have you done on your own design for the DCX?

In can try to use 100K and 1K and measure when I'll will have built a prototype.




Again a nice question.

We need to have the ADC/DAC run at maximum level for best performance. Input gain is for this really important. I'm not sure if the 1632 can manage the needed gain. May be a first buffer is needed.

BUT, the DEQ also have it's own internal RTA level meter screen with db or dbu print. If we enter, for example, a -40db signal, the screen must show -40db...

In the 1632 datasheet, they also write "The feedback resistor values (R3 and R4) should be kept relatively low, as indicated, for best noise performance."

As result of all these parameters to "manage", I really don't know how to select the right gain. Again, may be the prototype will help to answer all this. Or a guru of dbu. I'm so bad with math. ;)

Stef...


Ps : I made a mistake in the beta4 (AINR- and AIN+ are inverted). beta5 coming soon. ;)

Ps : what about THS4131?


The ocm will not take a lot, but the datasheet of the ref diode gives 0.6 ohms specs for 100uA current. To allow for some ocm input current, you could use 200uA from 15-2.5V = 62k series.

For the gain, I use x1 with equal input- and feedback resistors (2.2k), but I have gain stage in front for the level control which you don't need.

With 2.2k the input Z is higher (2.2k) and still low enough for good performance. Since the ADC needs about 2VRMS for full range, you can fully drive it with 2V at the input. If that is not enough gain, make your fb resistor 3.3k, the input resistor 1.5k and you have better than 1V input sensitivity. No need to wrestle with dB or something, it's much simpler than that. The 1632 is just like two opamps, and the gain of opamps is just Rfb/Rin (in inverting mode).

Don't worry about the extra stuff to isolate cap loads. The cap loads of the ADC are already isolated by the two 487 ohms series resistors on the DSP board.

Jan Didden
 
Ok, thanks.

If I summarize:

I can start to play with R1, R2, R3 R4 = 2K2

Using Rfb/Rin, I can calculate the gain (less 1/3 with the resistor divider at AK5393 input).

What is the influence of R5/R6 if not C3? Can I use 10R, 100R fro R5/R6?

Do you see an advantage to add C3 (as we already have 1.5nF on the dsp board)?

[IMGDEAD]http://www.awdiy.com/uploads/images/DEQ-Analog/opa1632_basic_circuit.jpg[/IMGDEAD]

Next week, I'll will be in vacations. I'll start to work on the pcb. It was difficult to find an HE10 female pcb connector. Currently, I only found one model from 3M in the 9100 serie (9140-4500PL).

[IMGDEAD]http://fr.farnell.com/productimages/farnell/standard/142832007-40.jpg[/IMGDEAD]