Help to build an AMP for SHP9500+4xTDA1543 chinese DAC

Shaddonai

Member
2019-10-31 4:09 pm
Hi all! I´m new o his forum. I recently bought an SHP9500, and I´m about to buy this DAC . Since I'm about to start school holidays, I think I might tackle on a little project and build the amp myself (I know about electronic, have basis on Altium Designer ,electronic theory , and I can print the PCB with ferric chloride) . The thing is, the 4xTDA1543 DAC says in it's description that it can output 2Vrms. Using this explanation (Designing headphone amplifier circuit, need help - Do It Yourself - PS Audio), I calculated that I would need 7,5 mW to push 110dB, so the amp should be able to give 0,5V, but I wanted to go for 0,8V for having some headroom. So, what can I do, since the voltage I need is lower than the output of the DAC itself without an amp (I thought of putting a resistor in serie, but forums says it ruins the frecuency response)
P.D.:I still don´t know if doign the AMP with transistors or OPAMPS. Wouldn´t mind some advice on there.
 

Ben Mah

Member
Paid Member
2012-05-18 6:36 pm
Vancouver
A volume potentiometer at the amplifier input will lower the input voltage.

If you want to build an amplifier that is already designed and has a board available, you can look at the Whammy or Noir, both have boards available at the diyAudio store, and there are threads on diyAudio.
 
Tell us what your headphones are then we can check your calculations for how much voltage you need. Certainly a TDA1543 DAC powered from 5V cannot output anywhere near to 2VRMS, it will be more like 2V p-p (0.7VRMS). With very sensitive headphones (IEMs perhaps) you won't need any voltage gain at all, just a unity gain buffer will do the job. You can always turn down the volume digitally if that sounds too loud still.

If you only need a buffer and not any voltage gain then you can use discrete transistors to build an SE classA buffer which is a pretty simple circuit.
 

Shaddonai

Member
2019-10-31 4:09 pm
Well, thank you very much for your answers! The headphones are the SHP9500, with a sensitivity of 98 mW/dB and 32 ohms of impedance. Another thing, wouln´t the potentiometer, which would be at the end a resistor in serie with the output, ruin the frecuency response. Also, I've lookjed upon this transistor-based AMP. Would it suit my needs?
 

Ben Mah

Member
Paid Member
2012-05-18 6:36 pm
Vancouver
I don't have any experience with either. However, both the Whammy and Noir have been built by quite a few people. You can read the threads for their experiences. The Whammy was designed by Wayne Colburn of Pass Labs and the Noir was designed by Mark Johnson. Both know their way around electronics. I'm sure both amplifiers are quite good. It's just a matter of preference.

Another class A headphone amp that will have boards available in the future is the Nelson Pass designed ACP+. I was fortunate to be able to build one at BAF2019 and I like it.
 

Shaddonai

Member
2019-10-31 4:09 pm
Thank you! I asked because I wanted to spend as little as posible, so that means I need to design the PCB and print it myself, and designing something like the Noir is at a higher level than the one I listed. Anyway, I was answered in other forum that, If i didn't need the voltage amplification, I could just put a buffer. Is that true?
 
I checked out the SHP9500 the Philips data says 'sensitivity 101dB' but not the stimulus for that. Could be 101dB for 1mW - if so then they can deliver 124dB at full tilt which seems plausible.

Given the 101dB/1mW spec they need 179mVRMS at this power level and 1.8VRMS takes them to 121dB. Your DAC only goes to 0.7VRMS so that will result in 113dB in absence of voltage gain, probably still plenty loud enough.

I have a classA buffer design which will work - the gerbers for the 5*5cm PCB are available too : lingDAC - cost effective RBCD multibit DAC design
 

Shaddonai

Member
2019-10-31 4:09 pm
Hey, that´s great. Such a coincidence, beforre discovering the chinese DAC, I wanted to build the phiDAC, that I think was a more sikmplistic version of the lingDAC (Saw it on a hackaday project, Is it the same). Also, the gerbers are in that thread.
You are helping me a lot, thank you!
 

Shaddonai

Member
2019-10-31 4:09 pm
Sorry, but I couldn't find the gerbers. The ones that I found were the phiDAC v1, and I couldn't find the buffer PCB. Could you link me where they are? Also, since I'm going to iron them to the copper PCB, do you have the files alltogether? Like, with all the holes and layers in one single file, in a .pcbdoc file . Thanks in advance.
 
Last edited:
Gerbers for the SE buffer are here : lingDAC - cost effective RBCD multibit DAC design
I am not sure what you mean by 'all together' ? Drill files and gerber files don't mix they are always separate. For a two layer board you don't want all the tracks in the same file as they'll get confused. You can view them together with a gerber viewing application though - KiCad has one as part of its package but I think there are stand-alone viewers.


I've never heard of 'pcbdoc' file before, I will do a search on it. Ah I see its an Altium file - these gerbers were created on EasyEDA, not Altium. I think EasyEDA may be able to export it. I will report back.
 
Last edited:

Shaddonai

Member
2019-10-31 4:09 pm
For the method I print PCBs, (Is something like this), I need all the layers together, including the drill locations. In Altium, then, i configure two differnet copies, one with the bottom layer and it´s stuff, and another one with the top layer and it´s stuff. The I print it with toner, iron it and let it rest on ferric chloride for about 10 minutes.
 
I watched the first half of the video - it seems you don't need Altium to create the printouts, a gerber viewing program will work just as well. While EasyEDA does have an export to Altium function they're not 100% sure it conveys all the details with complete accuracy.


So I suggest to reduce the possibility of errors you go the gerber viewer way. How to you propose to make the vias by the way? This board has many for stitching the top and bottom 'heatsink' areas together to achieve the lowest thermal resistance.
 
Well, in this method, to join bottom and top layer I just drill a hole and put a component cutten leg, and solder it on both. I think that will resolve the electrical issue, but the thermal... I don´t think it will work. Does attaching small heatsinks on the MOSFETs achieve the same thing?
 
I don't think the thermal issues are a problem - if you're using copper wires then the thermal resistance will be lower than for a via which only has copper down the sides. I was thinking more of the labour involved - there are around 20 vias on each 'heatsink' area, multiplied by 4 heatsinks which is a very long job. Compared to getting the PCBs made at a factory which (here at least, in China) costs about $10 including shipping for 40 PCBs.

The MOSFETs are SOT-23 package, I've never encountered a heatsink for that package size.
 

lasercut

Member
Paid Member
2017-12-19 4:35 pm
If you are up to the task of building above circuits then my recommendation would be ultra simple and inexpensive OPA1622 voltage follower.
Literally 1 component, just needs power supply... and it will work well with just about any power supply thanks to wide supply range and high PSRR.

Dont think there is some compromise with this amp, the sound quality is actually insane with this chip with tons of power, it could be built very compact and be battery powered if you wanted portability too....

You can get OPA1622 pre-soldered to DIP adapter on ebay, they usually come from china which can take a few weeks.
Although a bit pricier audiophonics sells them in europe, there could be US seller too... not sure.