TDA1541A Diy Pcb - "Distinction-1541 v2"

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Lovers of the TDA1541A -

"Distinction-1541 V1" Was a success thanks to the combined knowledge of the DIY community.

I start this thread for a place for myself to share what i'm working on with the TDA1541A, however unprofessional it may seem. I admit i'm certainly no expert. I just want to share my experiences with like minded people.

A new direction for the project has obviously been led by myself; wanting more from the TDA1541A - Especially after hearing just what V1 was capable of.

Design goals for v2:
-Simplicity
-Cost effective vs performance
-Attention to detail of Pcb design topology
-The use of the latest passive technology in SMD formats

The output stage for the TDA1541A is normally a personal choice for particular reasons so this part will be left out. I continue to use the "CEN Passive IV"

Attached to this post you will find a few pictures of the progress v2 has currently undergone. These pictures by no means represents what the finished product may or may not resemble - they simply show what i'm working on at the moment to attain the ultimate goal of the final product.

Please note that there are design compromises in every PCB design and this one is no different - all i'm doing is selecting the parts to the system I feel to be important based on simulation, practical testing and of course, most importantly - listening.

Ryan.


Distinction-1541 v2 - GB interest list. - diyAudio
 

Attachments

  • atten progress.jpg
    atten progress.jpg
    424.9 KB · Views: 4,457
  • v2 progress.jpg
    v2 progress.jpg
    907.2 KB · Views: 4,297
  • Different values1.jpg
    Different values1.jpg
    483.8 KB · Views: 4,086
Last edited:
Re-post of "zener string"

Thanks Walter.

Attached is a Simulation which has already been posted on v1 thread.

This is not the final schematics - just something i've been playing with.

Hopefully I'll have the time to build and experiment with this soon.

You will notice I've added inductance to simulate the PCB trace inductance - It's a bit crude but it does alter the results somewhat.
 

Attachments

  • zener string idea.jpg
    zener string idea.jpg
    431.9 KB · Views: 3,878
  • zener string idea z-out vs phase.jpg
    zener string idea z-out vs phase.jpg
    715.6 KB · Views: 3,674
Cen

This might help. Should be easy to do on vero board - which is what I did in the beginning. ;)

You could omit DAC -15, R8, and NCT 5k.

Jfets should be matched for Idss.
 

Attachments

  • Cen with k216 ccs.jpg
    Cen with k216 ccs.jpg
    57.4 KB · Views: 1,623
Last edited:
  • Like
Reactions: 1 user
Update

Ryan is your design finalised for this? If so can you post a BOM please.

Hi Luke,

Ill answer here to keep the two threads separate.

I'm still very much in the design phase; plenty of work ahead before a prototype is ready for testing.

Recently I've been working on laying out the attenuation circuit.

I've decided it may be a better solution to add a reclocker instead of a signal buffer (needed to correctly drive the atten circuit) as its only one extra part - SMB connector for the master clock.

Image attached - Please note this is still work in progress.
 

Attachments

  • updatev2.jpg
    updatev2.jpg
    270.6 KB · Views: 1,175
Sorry I have thread crossover confusion. It looks really nice, it seems to be getting very SMD based now, getting out of my comfort zone. However, don't let that stop you, I suspect through hole days have another 5 years or so to go so maybe a good time to get some practice in. Its nice to see the odd post to let us know what your up to, we need something to talk about and critique:) One request would be to keep the SMD parts to 1206, nothing microscopic if at all possible.
 
Update

Just a bit of an update of where i'm at with the new design.

As you can see from the image attached, all the components have now been placed but not yet routed, i'm taking great care in the attention to detail - As we know the 1541 is very fussy with layout.

Still much work before I get a design ready to prototype and test. Thanks for everyone's patience.

Next steps will be to calculate heat dissipation techniques and how big to make the copper pours - the only challenging one will be the -15V with ~1W of heat to deal with. The CCS for the 3 supplies is only around 500mW so should not be an issue, might have to physically separate them further apart.

Still a few areas in the schematics that need to be looked over, and also need to add ground planes and power planes.

:up:
 

Attachments

  • updatev2 2.jpg
    updatev2 2.jpg
    529 KB · Views: 1,468
Last edited:
Resistive attenuator simulation.

Hi All,

It has come to my attention in a discussion that; the diode attenuators may not be the best option for attenuation. Due to the large current pulse - Ground bounce will likely be an issue. Originally I was attracted to the fast rising edge that the diode attenuator produced in the hope that this will reduce trigger uncertainty within the 1541 thus less jitter.

So the compromise going from the diode attenuator to a resistive attenuator will be a slower rise time, but there will be less harmonics entering the 1541. Also there will only be a ~5mA pulse through the resistive attenuator as opposed to ~100mA pulse in the diode attenuator, so we will also be getting a lot less/zero bounce on the digital ground plane.

Attached is a simulation of what ill try in the next few days. I should be able to use the diode attenuator pcb instead of making a new one.

Ill report back with a few scope shots.

:up:
 

Attachments

  • resistive attenuator sim.jpg
    resistive attenuator sim.jpg
    140.1 KB · Views: 1,417
Last edited:
Resistive attenuator test.

Attached is a scope shot of the resistive attenuators in action. I didn't have all the exact values on hand thus why the DC bias is at about 1.3V with a total amplitude of about 2V. Probably should be aiming for a range not too different anyway as the input threshold for the 1541 is about 1.6V if i remember correctly.

If you look at the image you may notice that the trace is not too sharp compared to the diode attenuators, I think this might be due to having thick film resistors in the signal path contributing their large noise spectrum - which will translate to jitter. I'll replace them with some MELF and take another look.
 

Attachments

  • resistive attenuator test.jpg
    resistive attenuator test.jpg
    958.4 KB · Views: 1,328
MELF vs Thick Film

Hi Guys,

Attached is a comparison of MELF and thick film resistors in the resistive attenuator circuit posted above.

The shot on the left of each image is the Circuit with all melf resistors, and the right in both images is the shot I took yesterday with the thick film resistors.

Notice the width of the rising edge. Amazing what good quality resistors can do. Makes sense though - less noise, less jitter.

I haven't had a chance to have a good listen with this attenuator, but there are already advantages using this setup.
-I can drop the 5V reg on the distinction PCB (Originally there to power the diode attenuators and isolate the flip flop and buffer from the large current pulse the diode attenuator was producing) and run all the digital side off the one 3.3V shunt reg.
-Lower part count
-Smaller PCB (maybe)
-No ground bounce to worry about.
-Less high frequency harmonics entering the 1541.

Seems like a better option.
 

Attachments

  • resistive attenuator test melf comparison.jpg
    resistive attenuator test melf comparison.jpg
    563.8 KB · Views: 1,175
  • resistive attenuator test melf comparison closeup.jpg
    resistive attenuator test melf comparison closeup.jpg
    290.4 KB · Views: 1,090
Last edited:
Update

Preliminary design. Not yet tested or fabricated.

Hi Guys,

Time for a bit of an update.

Attached is a preliminary design layout of layers 1-4 which I have been working on.

I've added the whole voltage reg as indicated in the schematics in a previous post - so all you'll need is a pre regulated, pre filtered 24V supply, and a 6V supply for the digital side. More details once I get this all tested.

I've also added 2220 Pads to try out those Rubycon ST 22uF film caps - they look the goods, and i'm keen on trying these without any Electrolytic caps even though this may cause an impedance hump at 100 - 200k.

On the digital side - I've taken a lot of care in shielding - hence all the via stitching. All stiplines have been properly calculated for imdedance. Ground fill surrounding stiplines from layers 2 and 4. Power planes on layer 1 with Ground on layer 2 giving extra distributed PSU decoupling capacitance. Split Analog and Digital ground planes connected at pin 14.

I've designed in the resistive attenuators - definitely a better solution in terms of current flow/ground bounce compared to the diode attenuators - not that they didn't seem to perform well, just all that current flow will likely have caused issues.

Thermal pads have been calculated to around 25 degrees celsius above ambient (-15V) Once tested I'll know for sure how far the analog supplies can be pushed. With extra current shunted to gnd, performance is improved. I've added trimmers in parallel to the CCS current setting element to tune this.

A separate return path has been kept for the 14 decoupling caps.

A 3.3mH choke with a DC resistance of 9 ohms has been added in place of the CCS resistor on the analog side - as this gives extra ripple rejection from the source (3dB). If this was increased to 10mH - according to the simulated you would get around 9dB of extra ripple rejection.

I'm starting to feel comfortable with this design. Shouldn't be too long for a prototype to be ordered soon.

Regards,

Ryan
 

Attachments

  • UpdateSep.jpg
    UpdateSep.jpg
    285.6 KB · Views: 1,049
Close up of digital side.

Hi Guys,

Been doing a few final modifications before I get a prototype made.

Attached is what I have so far. Might be worth throwing in a bit more ground stitching.

The Image is looking from the bottom of the board, layer 2 is the goldish colour - Continuous ground plane (split for digital and analog and connected at pin 14).

Layer 3 is in green and contains the striplines.

Layer 4 is in blue - ground fill, components.

To the right of the image you will see the MCK input stripline sweep past the reclocker and continue to the MCK output buffer to maintain 50ohm impedance.

Ryan
 

Attachments

  • Digital close up.jpg
    Digital close up.jpg
    331.2 KB · Views: 453
Thanks Nick,

I'm really looking forward to it myself. It will be very interesting to see how these shunt regs perform so close to the pins compared to the salas shunt regs roughly 70mm away (in my setup). If all goes well in the prototype I have every intention to make it available. :up:
 
Thanks Nick,

I'm really looking forward to it myself. It will be very interesting to see how these shunt regs perform so close to the pins compared to the salas shunt regs roughly 70mm away (in my setup). If all goes well in the prototype I have every intention to make it available. :up:

The shunt regs can only sound good. Judging by your attempt to simulate trace inductance by inserting series inductors into power supply rails few posts back - I don't see any reason not to expect an implementation capable of producing excellent results.

Regards,
Nick
 
Hi Nick,

Yeah I have a feeling they will sound better than the salas shunts (great regs, hands down) purely based on the fact that they will be 3mm away from the pins, so the current loop is kept as small as possible. The power planes will also provide extra distributed capacitance - Every little contribution to performance will help. The simulations can only show half the story though, especially when It comes to audio. It shall be very interesting to hear the results.

Ryan
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.