Tiny TAS5558 +TAS5624 "Power_DAC"+SMPS

[IVX]

abraxalito, you know, my noise-shaper has no such hill in highs on the noise floor, so I tend to suspect something else.

 

[IVX]

Hi, seems noise-shaper is about ok now, THD 1kHz 4Ohm resistor .005% 1/2 max power i.e. 50W. Before filter coils THD the same for 1kHz and 5kHz, however, after coils THD rise 3 times at 5kHz. BTW, speaker load THD 10 times higher than resistor load )) The reason is very simple, the coil impedance at >3kHz is too high to damping non-sinusoidal currents of speakers. And again I recollect Mr. Westlake with his .0000*% THD, if the real loudspeaker will return THD to .05% why it was so exciting? Anyway, .05% it is Hi-Fi as well

 

[IVX]

I've decided to order diecast heatsink yet. Somethin like that:

 

[doctormord]

You did the math for the required R_th for the particular heatsink yet? (With/Without forced cooling, with/Without envlorue) Or just because it's nice?

 

[IVX]

Oh, is it nice? Thank you doctormord And no, I didn't any thermal analyze of that because the amp works fine with the smaller heatsink. Probably a good idea to try CNC dummy heatsink for proofing.

 

[IVX]

recently I've implemented a dual feedback loop, before and after the LC filter. I took the Ti idea from the SLAA788A–September 2017–Revised March 2018. I won't say if it helps regarding the THD on resistor load, rather doesn't(even worse, 1kHz -3db .015%, -20db .0025%). However, it helps to fit frequency response fluctuations into the 1db room(+/-.5db 4/8ohm), and also reduces the THD on the speaker i.e. real load. S/N about -107db(A) with noise-shaper, -101db(A)(-102db if AC power voltage less than 160VAC, there is my SMPS idle noise issue) without. I try to keep the noise-shaper as simple as possible just because of the tiny size of everything, so I use only one THS4521 per channel(3rd order loop filter is passive network C/R-C/R-C/R). I think it's would be an interesting platform for experimentation around a noise-shaper, if someone has a time and passion to build a nice one, higher order classic noise-shaper(1 OP integrator per 1 order).
PS: I've checked the cost of tooling + 500pcs die-cast cases, it's about $10,000. In addition, need some plastic tooling for the rear/bottom panel.

An externally hosted image should be here but it was not working when we last tested it.

 

[IVX]

what is true DIY? I keep working on the PWM modulation index limiter, actually, I can limit that index in the TAS5558 but in that case, amp loses about 25W. It is pretty strange but the bootstrap 100nF caps discharges surprisedly quick if switching is stopped, around 20uS or so, that creates loud crack sounds.

 

[doctormord]

Add some diodes for an additional power pass to the bs caps.

 

[IVX]

doctormord, ok but what about 20Hz clipping case? > 25mS and 100uf there? I gonna try this one today:

 

[IVX]

It seems the problem is OCP, which actually shows undocumented behavior. It's really looked like a bootstrap UVP issue and only happens on the high side. As I understand the TAS5342 should work in a current limiter mode 1mS and next latch until external reset. In my case 5342 does Off the high side for 200uS, after 100uS of intensive ringing due to current limit action. I gonna try to add 2 SMA diodes per each half-bridge output to clamp ringing to GND and to VCC.

 

[IVX]

Haha, I was right, 4xES1DW helps. Now max power in the closed loop (noise-shaper On) mode 95.6W@4ohm (THD+N = 1% 100W in open loop mode), however 1% of the THD is unreachable only .039%(THD+N = .067%), and next +.1db makes THD 20%. So I gonna add into the IOS/Android app controls two parameters more: Max modulation index for closed loop mode and second one for the open loop as well.

 

[IVX]

Did you guys see a funny YouTubers who upload videos captured with a microphone from some HiFi setup, to share sound feeling? It is too much fishy for me yet, so I made the record 24/96 Diana Krall CD track "Like Someone in Love" just from my speakers tied to HiFi-TOY Mk-1 amp (now I call so the first amp version tas5558->tas5624) at the moderate volume level. So the chain is: CD track->USB->CM6631A->tas5558->tas5624->600:600 iso-trafos->OPA1612->CS5381->CM6632->USB->GoogleDrive. Personally, I do not hear any difference vs original CD at all, what about you? I can give you a link only to my record to respect copyright, so I hope you have an original CD and may compare 5558.flac - Google Drive

 

[IVX]

HiFiTOY app GUI YouTube
Dynamically allocated biquads between 2/4ord filters and parametric EQs

 

[doctormord]

Pretty cool.

 

[IVX]

Thank you doc, many times I heard about this one GUI idea that it is an ideal filter interface i.e. no reason to improve that anymore, like a spoon(we can add a battery and WiFI into a spoon but it's no way to improve a spoon yet)

 

[doctormord]

I'd like to have something like that for a PCM5242 or 5252 as a frontend for amplifiers.

 

[IVX]

it's possible, and not so costly, however, you need some DSP to control.

 

[doctormord]

The DSP is in the PCM5242, the 5252 only has BiQuads and PGA.

 

[IVX]

aah, ok, got it. If you really want it we can help, for some money.

 

[lutkeveld]

I agree! The HybridFlow chips from TI need a GUI. Their chips are amazing but lack the ease of use of SigmaStudio and the like. Especially since you are forced to use their EVMs to set up the settings.