TDA1387 x8 DAC: let's check its design, mod it -or not-, play music -or not! :(-

At long last here's the schematic (hand drawn as LTSpice's TL431 symbol is really confusing) for the discrete I/V modification.

Implementing the balanced outputs requires four of the BC807s (or BC327s if you're doing this non-SMT) plus associated FBs and 1k1 resistors. Only one TL431, 220uF and set of voltage setting resistors is required though - there's plenty of current available to feed the 3.8V voltage reference to 4 I/V transistors.

<edit> Incidentally I've implemented a couple more lily-gilding mods to my current DAC. First I've replaced the resistors feeding the shunts from the res caps with LM317s configured as CCSs in series with resistors for ballast. This gave a subtle improvement in bass 'weight' or definition probably as a result of having better LF PSRR with the current source. These 317s have been included on all three of the analog supplies (but not the digital). I'm also experimenting with adding biassing current sources to the I/V transistors which I'll talk about if there's any interest in that.

Can i use this for single TDA1387?
 
Yeah you can though you'll probably want to increase the I/V resistor (currently 1k1) by a factor of 2 (if you're not using the passive filter) or 4 (if you plan a passive filter with 39mH inductors).

<edit> Btw matt - I recently came across this encouraging post over on Audiocircle - http://www.audiocircle.com/index.php?topic=134972.0 :D A few views but no replies - do those guys not realize the qualities of NOS DACs?
 
Last edited:
good

At long last here's the schematic (hand drawn as LTSpice's TL431 symbol is really confusing) for the discrete I/V modification.

Implementing the balanced outputs requires four of the BC807s (or BC327s if you're doing this non-SMT) plus associated FBs and 1k1 resistors. Only one TL431, 220uF and set of voltage setting resistors is required though - there's plenty of current available to feed the 3.8V voltage reference to 4 I/V transistors.

<edit> Incidentally I've implemented a couple more lily-gilding mods to my current DAC. First I've replaced the resistors feeding the shunts from the res caps with LM317s configured as CCSs in series with resistors for ballast. This gave a subtle improvement in bass 'weight' or definition probably as a result of having better LF PSRR with the current source. These 317s have been included on all three of the analog supplies (but not the digital). I'm also experimenting with adding biassing current sources to the I/V transistors which I'll talk about if there's any interest in that.

Very interesting your scheme you use the dac as a current source for emitter but
way you don’t use the reference pin ?:p
 
The reference pin isn't used as a reference for the transistor that's right. It doesn't give the correct voltage. To get the highest swing I've set the reference voltage (3.8V) for the transistors as high as possible within the output compliance spec of the 1387 (up to 4.5V on a 6V supply).
 
An alternative

The reference pin isn't used as a reference for the transistor that's right. It doesn't give the correct voltage. To get the highest swing I've set the reference voltage (3.8V) for the transistors as high as possible within the output compliance spec of the 1387 (up to 4.5V on a 6V supply).

A connection like a NPN BJT with the base connected to the reference voltage of your TDA1387 and a Resistor connected from emitter to the GND.
The output on emitter Vout Ref is lower by a one Vbe
 
Last edited:
Not quite ...

The reference pin isn't used as a reference for the transistor that's right. It doesn't give the correct voltage. To get the highest swing I've set the reference voltage (3.8V) for the transistors as high as possible within the output compliance spec of the 1387 (up to 4.5V on a 6V supply).
Uh... not quite... the datasheet value for Vref is optimal. Philips didn't just toss off a convenient value for the sake of what most manufs (SoundBlaster, etc.) would use. They OPTIMIZED the 1387/1545A for the pwr/voltage/current available via commercial regulators.
If you're getting highest performance for YOUR project at some ad hoc Vref of your choice, you're most likely SERIOUSLY compromised somewhere else in your DACs topology. Find n' fix those ... then come back to Vref PER DATASHEET SPECS.
 
Alternative

IV-stage.png This is a more clear circuit you can use the BC550 560 or the same smd BC850 860 :)
Output on your R2
 
Last edited:
This circuit doesn't suit my needs because the Vref for TDA1387 is roughly one sixth of the supply voltage. In this case that's 1V - after buffering with an NPN EF that becomes 0.4V meaning the I/V transistor goes into saturation unless its output swing is below 100mV or so. Far too low an output voltage range for my needs.
 
errata corrige

This circuit doesn't suit my needs because the Vref for TDA1387 is roughly one sixth of the supply voltage. In this case that's 1V - after buffering with an NPN EF that becomes 0.4V meaning the I/V transistor goes into saturation unless its output swing is below 100mV or so. Far too low an output voltage range for my needs.


Yes i have made a mistake this works well with TDA1543 and TDA1545 .:(
Ok the output of tda1387 is a strange output .....:eek:, now i have read well the data sheet...the I/V stage have reference to ground .
If you connect my stage to the gnd as reference and a -6V on emitter Resistor of the first NPN it can work ?
 
Last edited:
Member
Joined 2008
Paid Member
for opamp folks I advise the LT1028 with 330pF, stable (4 Dacs and 1.3k feedback R).

Do you happen to know which specific LT1028? LT1028AC or LT1028C? The datasheet on page 3 shows the two parts have some different specs. I'm not sure which specs matter for this application.

Also: buy them direct from Linear Technology, or roll the dice with AliExpress? I mean, are these opamps something that are likely to be faked?
 
Member
Joined 2004
Paid Member
Matt and abraxalito:

I've ordered the 8x TDA1387 DAC and am now assembling my mouser order (I've got 3 to 4 weeks before the DAC arrives). You have suggested replacing the analog stage decoupling caps (specifically, "The stock DAC has two electrolytic and four film capacitors for opamp decoupling. Remove these capacitors and replace with as many 35V rated caps as you can fit. Find the lowest ESR you can get (Rubycon ZLH; Nichicon PW,HW; NCC KY, KZE)"). What should the target capacitance be?

Many thanks,
Scott
 
Member
Joined 2004
Paid Member
abraxalito:

Thank you, once again, for your prompt response! I'll start hunting for small diameter, low ESR 35V caps, favoring those listed in the wiki.

One other question: at some point in this thread, someone inquired whether it was necessary to remove the 1uF "105" caps on the underside of the pcb. I don't recall anyone answering the question. If 1000uF caps are soldered to the TDA1387s on the top side, do the 105 caps really need to be removed?

Regards,
Scott
 
Member
Joined 2008
Paid Member
Abraxalito, is there a "formal" name for the turret boards you are using to assemble the additional circuits? I'm thinking the fancier "next gen" mods might be too complex to try to assemble point-to-point. I think doing like you with those boards might be an easier approach. But I can't seem to find anything on Mouser, but I'm probably just not using the right search terms.