Comparison between TDA7293 and LM1875

[richwalters]

To be honest I've spent many a night burning the midnight oil trying to optimise the THD on the LM1875.
As others have mentioned, the IN/OUT ground interconnection is inevitable when it came to PSU' cap grounds, esp when used in Stereo. The only solution I found is when using in pairs is to do a tighter layout than recommended and keep input screens well away from AC field sources.
Optimising the THD at high frequencies "went to pot" when a channel separation test was done at 10Khz and this was one of the reasons why I avoided using the chip in multiple bus systems.

It might seem silly but the same rule applied to twisting tube heater supplies just as well goes for preamps or so. This is why I'm so against the toroidal transformer in such systems, unless the flux density has been reduced.

However I'm not totally against the chip, for normal Hi Fi usage it does offer an excellent S/N ratio.

rich

[peufeu]

It seems there are only two solutions to the ground problem :

- unbalanced amps with one power transformer per amplifier
- multi channel amps sharing the same power supplies must be balanced (ie. there is no "ground")

...

[richwalters]

The first and obvious is that both in/output grounds have to connect somewhere localised within the 1875 layout, otherwise the chip will take off into HF oscillations. Remember gain B/W product = 5.5MHz and 8V/us slew rate.
Okay...I treated the o/p section and power supply decoup as per RF layout practices, i.e Hi uF caps geometrically close as poss to the +/-chip power entry, and the LS return "on the furthest away" from the input but again close to the common-zero of the +/-, if you follow this. I used 10,000uF for supply caps.
The AC 100hz charge pulses aren't as bad as the effects of a poor LS return can have on quality, esp if lo ohm speakers are used. The low impedance return on the common zero is essential and I used double sided PCB with the full ground plane be used for the common zero. Then the signal input ground be furthest away or otherside of the power section. For stereo, do the same, except I used end-to-end arrangement with signal inputs in the middle. (Butterfly layout).
If you simply lash this hi slew rate amp up with birds nesting, then the performance is miserable.

That's about as far as I take it........good luck; it makes tube circuits look so much easier without all these side effects.

rich

[Easyamp]

You should never mix power grounds with signal grounds and if you must to maintain stability then the device would be useless.

You need to put the signal ground closest to true earth ground or at best (2 wire) directly connected to center tap or a resistor cap softner and then to C.T.

It doesn't matter how much supply or local bypassing you have you always have spikes on the power supply rails and ground which will be fed into the input ground/input signal offset and just inject all that crap into the input and be horible if not seperated.

Chassis ground should be true earth ground through a 1 watt 5~10 ohm resistor if not directly coupled and then connected accordingly to the amplifier. If a 2 wire plug is used (uhg) then to C.T.

Not that it needs to be said but even in the data sheet it is stated very clreary do not mix signal ground with power. I observes low level / high level seperation on all amps I build.

Excuse the child like drawing it was just a quicky I made in MS paint.

[richwalters]

That's the snag...lift off the signal ground through a low ohm and up goes THD. Strange I know.......
I'm currently using this amp in < ideal conditions > successfully with much beefed up joint grounding, although the rule about the layout I mentioned previously is pretty good in keeping the input return well away from the o/p ground. The hum and noise isn't the problem.
The hum and spikes are well, well down as per spec, (in fact excellent) but it's the conflict with the THD that shows up on my THD analyser is difficult to optimise.
The LS o/p return needs to be placed very carefully w.r.t split supply zero between the power supply caps which need to be low ESR types for best performance. I haven't used this amp with a single supply.......perhaps this might be better bet in a view of optimising THD.
I don't agree with National's THD#freq figures.....but, however, lets say for normal Hi Fi applications.... it works a treat.....and I will leave it at that.......

rich

[Easyamp]

I have seen this r/c circuit implemented to smooth out low level spikes in discrete amps but have never tried it or really heard much about it and so I'm not quite sure the best opamp configuration to implement this. Have you ever used it or think it may help reduce spikes into sig gnd when using single ground?

Curious, are you using a cap like 100~300p across the inputs, this helps when seperating the grounds to absorb possible emi problems incurred by the extra ground plain.

I was just looking at the lm1875's schematic, and is it just me or is that the most over engineered 20 watt device you've seen.

Cheers

[peufeu]

Easyamp :

Why use the transformer center tap and not the PSU filtering caps center ? You'll get more pulses using the center tap... There is a shared path between charging pulses and speaker ground in your schematic...

Besides what's the reason for using only 1 bridge ? We can use 8 diodes ?

[GregGC]

This circuit will have ~6 db higher gain for the high frequencies (above f=1/6.3*500*50pf). I don't think it'll smooth out anything. In fact quite the oposite. Transients will be more noticable.
A cap between -in and +in would though.

Greg

[Easyamp]

A cap between -in and +in would though.

Isn't this the inputs?

Why use the transformer center tap and not the PSU filtering caps center

It was just an example of 2 ground system not an actuall schematic of anything and your right you would return the sig gnd to the filter center, which is usually only a few inches from the TF's CT. I suspect probably as much for easy and orderly layout then any more injected hum unless it long run from TF to filter.

[jez]

On the subject of lm1875....
I have used these at work to design an amp to accept a balanced input and drive the load in balanced (bridge) mode. The balanced input was something I had no choice in as it had to work with the companies current product range. It uses two lm1875 in parralel to give the required current for each "push-pull" half (4 lm1875 in total per channel). I had loads of problems making it stable but eventually managed this by slightly unusual means...,anyway the final results, as measured on our audio precision system two, were quite surprising, over 100 watts RMS into 4R and THD about 0.005% !(1KHz and about 25 W).
Sorry,I have no idea what it sounds like as it was for industrial use and I never got to here it !, maybe I'll find the time to make another one for my own use someday but dont' hold your breath.
BTW, whats a gainclone (just joined the site and I'm probably the only one who does'nt know)