What’s On the Bench Tonight (OBT)

Hi Folks,

Wishing everyone a Merry Christmas! Hope you have time to spend with family and that everyone has safe travels if going anywhere.

Member @tiverson has generously made available, STL files for the mini FH3 inspired speaker that I originally made from foam core. He printed the whole thing and it looks and sounds great.


1735052971389.jpeg


Tongue and groove assembly joint:
1735053064260.jpeg


Listen here - 1.5W SET amp sounds amazing on this tiny speaker driver.
 
  • Like
Reactions: dawa and twocents
Happy New Year everyone! Wishing you all a great beginning to another 365 days of our lives and may you have time and opportunity to do a lot of DIY. 🎉🍾🎉🥂

Thank you to all of you who support my shop and this forum - it would not be as interesting or fun without your input.

On the bench is a TPA3255 amp that a customer sent back to me for repair. It has a left channel that sounds softer than the right. It was intermittent and now fully soft. It appears that the B channel half of the bridge output is flat lined. Debugging it now. It plays music and sounds good, just half the volume due to missing half the output swing.

I installed it into my test bed chassis and listening and probing around.
1735737270434.jpeg


1735737284998.jpeg
 
I found the problem. A very rare case of a passive 0805 chip resistor in signal path before the opamp going open circuit. What would cause it to do that except massive over current? But it did not look burnt. I replaced it and all works now. Took a while to trace it part by part with a DMM and microscope, then verified with O-scope in operation.

Since I had my test rig out and my own amp board in the open, I decided to implement member Bucksbunny’s capacitive post filter feedback for reducing overshoot. It requires testing with a square wave to select the right value circa 100pF NP0. Basically connect output at speaker back to TPA3255 signal inputs with a series 100pF capacitor.

I am at 66pF now (using 33pF 2500v NP0 1210 caps stacked in parallel). Adding another will bring to 99pF.

Scope shows a mild small overshoot.

1735760392131.jpeg

Listening tests now.
 
  • Like
Reactions: Plott
On the bench tonight, I moved the location of the caps to the input side to decouple the lengths of feedback wires from becoming an antenna. Also changed to 220pF 100v 0603 C0G.

1735887124955.jpeg


Measurement setup with about 22Vpp into 5ohm load. With or without load the output looks the same on Oscope.

1735887181645.jpeg


Squarewave looks like this with a mild overshoot and about 4 cycles of damped ringing.

1735887226890.jpeg


Listening to it so far it sounds great. I can’t say for sure but “smoother” is how I would describe it. I put it back into main amp rack for li g term listening.
 
  • Like
Reactions: uncola
On the bench is an old foam core HyperCube speaker cabinet that I repainted with black and silver paint and installed some new PC83-8 drivers. I measured the raw signal and developed a passive filter for it.
1736109725644.jpeg

Filter in Xsim:
1736109749561.png

Green is raw:
1736109767420.png

Harmonic distortion:
1736109789669.jpeg

Wago P2P Filter:
1736109885226.jpeg


Recording made with Caldera 12 sub. Amp is TPA3255 modified as shown above:
 
Last edited:
  • Like
Reactions: Plott
On the bench is a comparison of the external filter with proper arrangement to reduce cross talk between the inductors and compact P2P soldered filter with the two air cores very close to each other but with orthogonal axis. The response is about the same with a small variation at 550Hz but I am not sure that is caused by the inductors as measured on different days and could be slight changes to external reflections etc. Harmonic distortion is about the same as well. I will let this stay as is then.

HyperCube-Internal-vs-External-P2P-Freq.jpg


Internal Filter:
HyperCube-Internal-HD.jpg


External Filter:
HyperCube-External-HD.jpg
 
  • Like
Reactions: jordheis
Love it: Say-to-Do Ratio. Being a mechanical engineer I like non dimensional numbers. But the acronym is too close to the other acronym STD. 😂

But SDR works well. Say-Do-Ratio.

Not to be confused with the Systems Definition Review part of lifecycle analysis.

But Do to Say Ratio might be better as you want that to be large. DSR.
 
Last edited:
  • Like
Reactions: von Ah and jordheis
Just adding power indicator LEDs like in my previous post in this thread. Some of my amps didn’t have any lights inside, so I decided where to add a resistor and an LED. The Mod86 got one added to the aux supply of its SMPS yesterday, and today the class A Boule amp got an LED added to the front end supply for each channel. A little bling while they sing. :yes:
 
Last edited:
  • Like
Reactions: uncola and xrk971
It’s in the HyperCube thread. You need to cut 3 shapes: rhombus, square, triangle. The dimensions and angles are set by the edge length of the square (the baffle where you mount the driver).


With foamcore or cardboard I cut straight square edges. But if using thick wood, you need to bevel for flush fit and the angles are 60 deg for the bevels.

These images were made by Mudjester:

1738939181890.png

1738939214432.png


Tesserax said the dimensions are as follows based on S, the length of the side of the square:

Once you have decided what your square side length S will be,

Rhombus side length = S x .866

Triangle side lengths are S, (.866)S, (.866)S

The theory for the rhombic dodecahedron is here:

https://www.kjmaclean.com/Geometry/rhombicdodeca.html

Have fun! It’s pretty easy to make with foamcore and hot glue.