Fasten seat belts. TDA8932 pessimistic review.

Looked at eBay last night, nothing quite the same as the winding machine I bought here. The one I can almost recommend is this one : Coil Manual Hand Winding Machine Chuck Coil Winder Electric Dual Purpose | eBay

The reason it doesn't quite make my recommendation is because its not quite heavy enough - it needs to be bolted down to a bench not to wobble around in use. I have one of these as my second machine and haven't used it for that reason - my first machine also isn't quite heavy enough but at least its top is flat so I've put some weights on top which do the trick. This way I don't need to bolt it to anything.

Most winding machines on eBay don't seem to have a chuck, this means they're fine for larger diameter cores but probably EP17s won't fit over the spindle.
 
Here's a highly zoomed-in shot of one of my modded boards.

Points of note - I'm running twisted pair differential input from the EP17 transformer, there are 11k resistors in series to attenuate HF pickup. Two extra caps added for decoupling, one to the analog power supply (pin8) and one on the analog Vref (pin12). Pin12 goes to the trafo CT allowing deletion of the input caps. Pin10 (resistor osc frequency setting) is 0R to GND because I'm using an external oscillator.
553389d1465176954-fasten-seat-belts-tda8932-pessimistic-review-p60606-093148.jpg
Could you mention what modifications and purpose and pin numbers the added parts are attached to in the photo? I'm very curious!
It is okay to get wordy.
 
Whilst I welcome questions, I'm not really sure how to answer yours Daniel because I'm not clear on the level of explanation you're looking for.

At the level of listening, transformers on the inputs reduce the perceived (subjective) noise floor from the amps. Not in the sense of how much hiss they produce but rather in terms of the ability to hear into a recording through them. The soundstage deepens when transformers are included at the inputs, that's been my consistent experience. The sound somehow is less 'mechanical' (or if you prefer, more 'organic').

As regards your last post, I mentioned which pins the caps are connected to in the text you've already included with the picture. If any of that's unclear, do please ask for specific clarification.
 
The main reason is the work done by Brockbank and Wass. They showed mathematically that when listening to normal music, harmonic distortion is insignificant compared to IMD. That's because with music there's a vast number (at least hundreds) of discrete tones but each is at a relatively low level - under such conditions non-linearity results in IMD dominating over THD by orders of magnitude.

The second reason is that your hypothesis doesn't fit with the observation about the soundstage - that's a perception related to the background, when signal levels are really low. Transformers have significant distortion only at higher signal levels, and at lower frequencies.

FWIW my hypothesis for why trafos sound better in the input circuit is they're reducing the pickup on input cables, particularly HF. They have decent common-mode rejection properties. I've already ascertained that the input stage of the TDA8932 is rather susceptible to HF pickup, which is why I've included the 11k series resistors in my mods.
 
Last edited:
But doesn't the common-mode rejection directly correlate with the symmetry of the transformer?

Perhaps it does, I've not looked into it. But here we're comparing CMRR with the trafo, to without it. Given the apparent sensitivity of the input stage to HF ingress, there's probably more to it than just measurements of CMRR here.

And why does the soundstage is getting "deeper" or why can someone hear "deeper" I think the recording?
Psychoacoustics - lower the noise levels (confusing signals) the more the brain can decode the acoustic space on the recording. Have you come across Bregman's book 'Auditory Scene Analysis' by any chance? It sets out to understand how the brain works in auditory perception.


Incidentally the design part of the trafo thread's done now, what would you have me write about the process of winding? Feedback welcomed.
 
Last edited:
Yes, quite possible though it would be bridged rather than paralleled, then the transformer can adapt the impedance of the speaker to suit the amp. Something over 16ohms should mean it wouldn't overheat, though probably would need a stick-on heatsink.
However, what I would like is a mono parallel se TDA8932 amp.
It doesn't do incredibly high end (at the treble) when bridged, so I won't be doing that. I have mostly series-hybrid (devore) crossovers, and it costs too much to adapt them for bi-amp. So, no bridge amps for me! It is for sure that I am in need of driving the speakers that I already own.
I've heard how that chip sounds SE without the paralleling, and as you can imagine, then that is what made me curious what it would sound like with double the output linearity quality (paralleling). Also, parallel amplifiers have, generally, a prettier tone and a bigger bass dynamic wallop. So, you know that I'm terribly curious to experience that on an amplifier that has already done both of those categories fairly well. What if I enjoyed it? What if it raised the bar even farther on what we should expect? What if it defeats the need of extra transformers? I claim that it is both possible and likely for all of those capacities to exist simultaneously. However, I'm a bit short on clues for how to arrange it.
 
The treble's a problem when the load impedance is too low - since I've put the step-down trafo on the output, I have had no issues with it. So I reckon you'll be fine with the treble in bridged with a transformer to raise the impedance the amp sees.

To run in parallel rather than bridged is possible, but would take two transformers, or alternatively very careful gain matching and two primary windings on a single transformer.
 
The treble's a problem when the load impedance is too low - since I've put the step-down trafo on the output, I have had no issues with it. So I reckon you'll be fine with the treble in bridged with a transformer to raise the impedance the amp sees. To run in parallel rather than bridged is possible, but would take two transformers, or alternatively very careful gain matching and two primary windings on a single transformer.
It also means the SE mode TDA8932 as a Parallel amp, is not likely to need any audio signal transformer.

However, there may be output power and expenditure differences.
 
Founder of XSA-Labs
Joined 2012
Paid Member
Perhaps it does, I've not looked into it. But here we're comparing CMRR with the trafo, to without it. Given the apparent sensitivity of the input stage to HF ingress, there's probably more to it than just measurements of CMRR here.

Psychoacoustics - lower the noise levels (confusing signals) the more the brain can decode the acoustic space on the recording. Have you come across Bregman's book 'Auditory Scene Analysis' by any chance? It sets out to understand how the brain works in auditory perception.


Incidentally the design part of the trafo thread's done now, what would you have me write about the process of winding? Feedback welcomed.

Yes, please show how to actually wind.
Thanks,
X