TDA7293 Parallel kit from ebay (modular/slave style, no lossy emitter resistors)

Steve, you could build a parallel LM1875 and then you'll know. It is not easily done, but it is rather amazing.

my old parallel LM1875 setup came together with very little drama actually, possibly because i put together a few more single chip p-2-p units than i needed and matched up the most alike before tying them together parallel and end-to-end balanced with the mu-metal transformer on the input side. it was quite like a lower-power cousin of a Jeff Rowland 10.
the only real drama was when, as i pretty much expected, that trafo rolled off in the deep bass unless the linestage presented a low output Z. so that's why i ended up doing that parallel 6922 mu-follower bit. parafeed kept the trafo from saturating. a mosfet follower (as you would see in any recent CJ 6922 linestage) would also have worked, and perhaps even better than the mu follower. Coincidentally, CJ's chief engineer said as much to me last night when we crossed paths at a local restaurant.
 
Last edited:
As for imaging with your TDA7293// amp that's running 25vdc rails.

With the parallel amp there may or may not be some advantage to using ballast resistors.

Similar to MJ15003 on a 40W amp or MJ21194 on a 40W amp. . . The parallel amplifier's 90W linear capacity and doubled output device capacitance/noise, might be unnecessary for usage at 22W, so that a solo TDA7293 might possibly perform a little better.

There's also Edmund/Cosine style compensation to explore.
 
As for imaging with your TDA7293// amp that's running 25vdc rails.

With the parallel amp there may or may not be some advantage to using ballast resistors.

Similar to MJ15003 on a 40W amp or MJ21194 on a 40W amp. . . The parallel amplifier's 90W linear capacity and doubled output device capacitance/noise, might be unnecessary for usage at 22W, so that a solo TDA7293 might possibly perform a little better.

There's also Edmund/Cosine style compensation to explore.

since, as i mentioned, i'm going toward end-to-end balanced architecture, the output will be near 100 watts; and given a 6 ohm output load each side will "see" a 3 ohm load.
BTW, a one-chip board from the same designer (with the same Chinese symbols) is available on ebay.
 
The maximum transient current seen by the bridged amplifier will approach 17Apk.
Two chips in parallel may be able to meet this demand. I would recommend 3amplifiers.
No I won't. I recommend you use a different topology. Try the lme49811 with just one pair of 200W devices. That will meet your target of 100W into 6ohms and you won't need to bridge.
 
for anyone reading later, the case here is bridged, +/-25v regulated PSU, 6 ohm load. so P-P V approaches 25 x 2 x2 =100V. at 6 ohms I peak is about 17A. it's actually a bit less since the chip does not swing rail-to-rail but close enough allowing for the no-so-little matter of safety.
 
The speaker won't mind that the last link (cable) is not a balanced impedance connection.

ok so the suggestion is use a better chip and combine the two XLR inputs inside the box that houses that better chip.
a safe solution for this approach would be to sum one half the XLR input with the result of a buffered then inverted signal from the other half. (with the buffer keeping the inverting element's entire loop inside the case). or perhaps dispense with the inverted and feed the buffer out directly to the -IN of the PA.
 
Last edited:
Slightly off topics but I would like to use this parallel configuration for a bass guitar combo amp. I have a 300 VA 25-0-25 transformer. I am happy to accept 1% THD + N so what should I do paper from building the datasheet modular amp?

you'll likely get different answers from different people here but i'll tell you how i would "skin this bass cat" if it were mine. except that i will assume you don't want to switch to a buffered inverted topology which would require a lot of tinkering.
1- because heat will be a major issue i'd either use the three-chip version of the board (same designer) or better yet use another topology entirely (how many commercial bass amps use the TDA7293? use any chipamp?) or use 3 7293 all in slave mode and drive them with something other than the onboard driver -- TI makes some very good driver-suitable power op amps and Peter Millet sells an excellent board for them on ebay. if you do use the onboard drive then put its feedback network at the chip pins and use a very much larger blocking cap (EX: 220 uf paralleled with a .1uf wima or vishay film -- like the power decoupler -- seebelow)
2- i would not use any of the electrolytic capacitors supplied with the board. the values and quality are not suitable, especially for bass. for decoupling caps on V+ and V- i'd put them under the board at the chips power pins and use a panasonic FC 220uf 63volt paralleled with .1uf 63 to 100volt (a kemet x7r or a wima film or vishay film). for the bootstrap cap i'd use 220uf panasonic FC or the equivalent nichicon. for standby and mute caps any 10 to 47uf 50volt nichicon is good enough and cheap.
3- the box cap on the input is too low in value and quality. if you want an inexpensive solution, the panasonic 3.3uf film (ECQ-E2335KF) is quite decent though you may want to parallel 2 of them. a bit more expensive is the 6.8uf solen that parts express sells. any of these is too big for the board so install it off the board.
4- you'll need a real power supply not like the sort of stuff you see in some"gainclones".
5- you'll need a very large heat sink for reasonable durability.
6- for reasonable durability you'll probably need a better output compensation network not just the simple snubber on the board.
7- the speaker return (gnd) should go the the power ground star (midpoint of the two 220uf power decoupling caps). the ground line from the PSU should tie to that star. tie the star to the board power ground if you have put the power decoupling caps under the board. there are some other grounding scheme improvements that could also be done but this is the minimum in my view.
8- if you play in "dirty" (high EMI) places then the 220pf cap isn't up to it by itself. i'd use 1 or 2 43-series ferrite beads at the input cap.
7- use authentic 7293 (Mouser or Digikey or Newark) not anything from ebay. most of those ebay parts are "culls" that may be fairly easy to blow up and are not very stable compared to the real thing.
 
Last edited:
Thanks for the advice. I have a nice 300VA 25-0-25 transformer doing nothing at the moment and that is why I was looking at the 7293. I could build one of these in no time and the parallel option was the thing I liked.

Has anyone built the High Efficiency Version? That looks like a Class H implementation to me.

Ironically a bass amp does not need to go too low, in fact the area below 35Hz is a bad place for bass guitar speakers ( depending on the cabinet design).

This chip has been used by several bass and guitar amp makers including Marshall, Laney and Carlsbro to name a few.

However I am open to suggestions and am also looking at some Class D options,


Sent from my iPhone using Tapatalk
 
oh,yes, i was thinking 5-string bass, the lowest note of which is 32hz.
in your case the low end is easier to fix ie:
the replacement for the crappy input box cap can be a single 2.2uf or 3.3 uf film.
the bootstrap cap can be 56uf nichicon or better and preferably paralleled with .1uf film.
you could look at the Marshall, Laney and/or Carlsbro to get a feel for the PSU and heatsink requirements and whether they use inverting or non-inverting topology.

what is the available drive level? (this bard expect circa 2 VRMS) and what are you driving (EX:4x12 series parallel 8 ohm or what)?

if you are open to LM3886 i have a method to use 1 or 2 of the YuanJing 3 x LM3866 inverting board very effectively by using a buffer (to keep the entire NFB loop confined to the board) and slightly altering the feedback circuit (with a T-network). it is about as good as 150-300W solid state chipamps get in this price and complexity range. the inverting topo lowers the otherwise very ugly and non-musical 3rd harmonic distortion and its intermod by-products that can muddy up the bass.
 
Last edited:
I have some spare LM3866 chips so that my be a good start. Can you send me the details? Shall I PM you with my email address? I found this YJ bare PCB on ebay is this OK?

http://pages.ebay.com/link/?nav=item.view&alt=web&id=320744363105


Even with 5 string bass, too much fundamental at Low B/32Hz can muffle the overall sound but you have guessed correctly, I only play 4 string.


Sent from my iPhone using Tapatalk
 
Last edited:
I have some spare LM3866 chips so that my be a good start. Can you send me the details?

...

http://pages.ebay.com/link/?nav=item.view&alt=web&id=320744363105

that's the board. i get mine from grace lee -- at auction they usually go for about USD $4 including shipping.
if you go that route you can use 1 board (3 parallel LM3886), which would be safe with your trafo at 4 ohm load and provide about 120 WRMS. at 8 ohms load your trafo would be safe for 2 of these boards bridged (my current setup BTW) for about 240 WRMS.
i buffer the input using unity gain NJM4556 if capacitively coupled or OPA2227 if DC coupled.
if you bridge them you'll need a bridge adapter (i'm end-to-end balanced so i don't need one). for that you can add an inverting unity gain op amp after the buffer and before one of the boards, driving the other board directly from the buffer. theoretically it's a slightly ugly approach but it does work in the real world.
 
NFB T-net:
in place of the usual feedback series R, create a voltage divider from SigOut to SigGnd and at the midpoint connect the feedback series R. use 22k1 1% or .1% for that R and for the input series R. choose the voltage divider values to set the gain.
this makes for a quieter amp and also raises the input impedance to over 7k that you can use a 6.8uf Solen (the board calls for 4.7uf, which is marginal at the original design input impedance of 3.3k)