High performance 'chip' amplifier.

[Crackie]

I've been working on a pair of active speakers for the past few months.............whilst reviewing suitable amplifiers for the project I came across an amp which might be of interest to some PHers.

The main power amp chip, called the TPA3251, is produced by Texas Instruments; the specs are incredible. When fed with a suitable ( 350+ watt ) power supply. The chip is capable of supplying a genuine 2 x 175w per channel and has been optimised to work into 4 ohm loads.

When fed with a 36v 10amp supply https://www.amazon.co.uk/gp/product/B07D6RJQYJ/ref... the following amp is capable of delivering remarkable power outputs, particularly into difficult loads https://www.amazon.co.uk/gp/product/B08FR1F9LC/ref...

The Aiyima A04 amp, which uses a circuit closely based upon Texas Instruments' application note circuit, delivers a signal to noise ration of 111dB and a dynamic range of 115dB whilst also delivering incredible out power into low impedance loads. The figures below are when the 36V power supply voltage is turned down to 32V. This is because the Aiyima A04's main power supply caps are rated at 35V.

2 x 65w into 8 ohms
2 x 120w into 4 ohms
2 x 150w into 3 ohms.

Here's TI's data sheet for the TPA3251 chip https://www.ti.com/lit/ds/symlink/tpa3251.pdf

The thread below is from the DIYAudio forum where this amp's performance has caused quite a stir. The standard £74.00 amp ships with a 32v 5amp power supply; when used this way it will still deliver 2 x 65w into 8 ohms but will be limited to 2 x 80w into 4 ohms. In order to deliver the higher powers listed above, into difficult low impedance speakers, the 350w supply, linked above, is needed..........

https://www.diyaudio.com/forums/class-d/362341-aiy...

Anyway, hope this is of interest........should be relatively easy to justify the man maths and treat yourself to an A04 'Present to self'.

Edited by Crackie on Tuesday 7th December 11:09

[TonyRPH]

These little amps are certainly capable of a lot of power, however I've yet to hear one that didn't offend me in the treble region.

I've not tried this particular chipset though.

[Crackie]

TonyRPH said:

These little amps are certainly capable of a lot of power, however I've yet to hear one that didn't offend me in the treble region.

I've not tried this particular chipset though.

I agree that treble quality on the earlier class D and Tripath amps wasn't great during gen1 and gen2..........some of the Gen3 chips were fine ( Texas Instruments TAS5630 and the Analog Devices ADAU1592 ). I think Temple Audio still use, and iirc B&W have used, the AD ADAU 1592.

I've measured and listened to many over the years; things have improved enormously with the most recent generations..............much depends upon the quality of the LC circuit in the output stage. The quality of the stock 'off the shelf' output inductors has changed for the better in recent years and the gen 4 ( Texas Instruments TPA3116D2 is good imho ) and 5 chipset quality have improved enormously too.

The Aiyima uses NE5532 op amps which are plugged into sockets rather than soldered; this allows users to easily swap them out and try alternatives such as the OPA2604 or OPA2134. Many credit the 2134 with replacing the 5532's etched detail with a valve type smoothness for example; not tried this myself yet though.

I've just bought a pair of the Aiyima's with a pair of the 350W PSUs. I've got one of these https://www.daytonaudio.com/product/1551/dsp-408-4... taking care of crossover/response correction/time alignment/equalisation duties on the active speaker project. Will report back when I've had a play with the amps.

You mentioned the chips being able to deliver a lot of power..........here is the data sheet for the 'sister' TPA3255 chip. The output levels, when fed with a suitably beefy 50-54V supply, and appropriate heatsink, are remarkable. 2 x 260W into 4 ohms when configured for bridge tied load and 1 x 480 into 2 ohms when used in PBTL mode ( Parallel Bridge Tied Load ). https://www.ti.com/lit/ds/symlink/tpa3255.pdf?ts=1...

Edited by Crackie on Tuesday 7th December 08:49

[TonyRPH]

Crackie said:

I agree that treble quality on the earlier class D and Tripath amps wasn't great during gen1 and gen2..........some of the Gen3 chips were fine ( Texas Instruments TAS5630 and the Analog Devices ADAU1592 ). I think Temple Audio still use and iirc B&W have used the AD ADAU 1592.

I've measured and listened to many over the years; things have improved enormously with the most recent generations..............much depends upon the quality of the LC circuit in the output stage. The quality of the stock 'off the shelf' output inductors has changed for the better recently and the gen 4 and 5 chip quality has improved enormously too.

Yes I agree on the inductors, which will almost certainly be responsible for the load dependent frequency response often see with Class D amps too.

I picked one up recently as part of a junk bundle I bought - it's quite a hefty board based on a TDA7498 - one of these and to be fair it didn't sound too bad when I tried it briefly (only to see if it worked!) - it's languished in a box ever since...

Crackie said:

The Aiyima uses NE5532 op amps; they are plugged into sockets rather than soldered. This allows users to easily swap out the NE5532 OP amps for alternatives such as the OPA2604 or OPA2134 . Many credit the 2134 with replacing the 5532 etched detail with a valve type smoothness for example; not tried this myself yet though.

The OPA2604 is obsolete. The OPA2134 is totally unsuited to this application as it ideally needs to 'see' a balanced source impedance for lowest noise and CMRR.

And besides the above - rolling OPAMPS does not change the sound lol.

Crackie said:

I've just bought a pair of the Aiyima's with a pair of the 350W PSUs. I've got one of these https://www.daytonaudio.com/product/1551/dsp-408-4... taking care of crossover duties on the active speaker project. Will report back when I've had a play.

You mentioned the chips being able to deliver a lot of power..........here is the data sheet for the 'sister' TPA3255 chip. The output levels, when fed with a suitably beefy 50-54V supply, and appropriate heatsink, are remarkable. https://www.ti.com/lit/ds/symlink/tpa3255.pdf?ts=1...

Edited by Crackie on Monday 6th December 23:31

The Aiyima A07 has the TPA3255. It received a fairly favourable review from Amir @ASR

[Crackie]

TonyRPH said:

Crackie said:

I agree that treble quality on the earlier class D and Tripath amps wasn't great during gen1 and gen2..........some of the Gen3 chips were fine ( Texas Instruments TAS5630 and the Analog Devices ADAU1592 ). I think Temple Audio still use and iirc B&W have used the AD ADAU 1592.

I've measured and listened to many over the years; things have improved enormously with the most recent generations..............much depends upon the quality of the LC circuit in the output stage. The quality of the stock 'off the shelf' output inductors has changed for the better recently and the gen 4 and 5 chip quality has improved enormously too.

Yes I agree on the inductors, which will almost certainly be responsible for the load dependent frequency response often see with Class D amps too.

I picked one up recently as part of a junk bundle I bought - it's quite a hefty board based on a TDA7498 - one of these and to be fair it didn't sound too bad when I tried it briefly (only to see if it worked!) - it's languished in a box ever since...

Crackie said:

The Aiyima uses NE5532 op amps; they are plugged into sockets rather than soldered. This allows users to easily swap out the NE5532 OP amps for alternatives such as the OPA2604 or OPA2134 . Many credit the 2134 with replacing the 5532 etched detail with a valve type smoothness for example; not tried this myself yet though.

The OPA2604 is obsolete. The OPA2134 is totally unsuited to this application as it ideally needs to 'see' a balanced source impedance for lowest noise and CMRR.

And besides the above - rolling OPAMPS does not change the sound lol.

Crackie said:

I've just bought a pair of the Aiyima's with a pair of the 350W PSUs. I've got one of these https://www.daytonaudio.com/product/1551/dsp-408-4... taking care of crossover duties on the active speaker project. Will report back when I've had a play.

You mentioned the chips being able to deliver a lot of power..........here is the data sheet for the 'sister' TPA3255 chip. The output levels, when fed with a suitably beefy 50-54V supply, and appropriate heatsink, are remarkable. https://www.ti.com/lit/ds/symlink/tpa3255.pdf?ts=1...

Edited by Crackie on Monday 6th December 23:31

The Aiyima A07 has the TPA3255. It received a fairly favourable review from Amir @ASR

The arguments around swapping op amps will has been raging for years. I've only done it once, in 1985 on my Philips 104 CD player. Replaced the NE5532s with AD712s; made not one iota of difference that I could tell...........

I've only received the amps yesterday and looked at the equivalence specs Burr Browns yet...........as mentioned above, I've not replaced the op amps myself. NE5532/5534 are already excellent in that application.

http://dicks-website.eu/noisecalculator/index.html

Edited by Crackie on Tuesday 7th December 10:33

[TonyRPH]

Crackie said:

The arguments around swapping op amps will has been raging for years. I've only done it once, in 1985 on my Philips 104 CD player. Replaced the NE5532s with AD712s; made not one iota of difference that I could tell...........

Likewise, I've not been able to hear any difference. In some cases measurable differences exist (as one would expect!) but these differences are inaudible.

Crackie said:

I've only received the amps yesterday and looked at the equivalence specs Burr Browns yet...........as mentioned above, I've not replaced the op amps myself. NE5532/5534 are already excellent in that application.

http://dicks-website.eu/noisecalculator/index.html

I recently modified a Cambridge 851D DAC (not through choice, it had developed a nasty fault, common to the model and it's variants). Rather than spend £££ on the original surface mount opamps, I designed a PCB to use the I/V converter specified by the DAC manufacturer (the Cambridge has dual AD1955 DACs).

I initially used an NE5534 as the I/V converter, but found that the OPA2134 OPA134 was actually some 3dB+ quieter (despite having an inferior noise figure on paper) and this was attributed to the matched source impedance from the DAC chip combined with the higher input currents on the 5534 (the OPA is a FET opamp hence low input noise currents).

Dicks Website is interesting, although there's no mention of CMRR or the effects of an unbalanced* source impedance on input noise.

• meaning the inputs of the opamp being unbalanced - not a balanced connection.

ETA: Should have put OPA134 not 2134! (the 134 is a single opamp - 2134 dual).

Edited by TonyRPH on Tuesday 7th December 10:52

[Crackie]

TonyRPH said:

Crackie said:

The arguments around swapping op amps will has been raging for years. I've only done it once, in 1985 on my Philips 104 CD player. Replaced the NE5532s with AD712s; made not one iota of difference that I could tell...........

Likewise, I've not been able to hear any difference. In some cases measurable differences exist (as one would expect!) but these differences are inaudible.

Crackie said:

I've only received the amps yesterday and looked at the equivalence specs Burr Browns yet...........as mentioned above, I've not replaced the op amps myself. NE5532/5534 are already excellent in that application.

http://dicks-website.eu/noisecalculator/index.html

I recently modified a Cambridge 851D DAC (not through choice, it had developed a nasty fault, common to the model and it's variants). Rather than spend £££ on the original surface mount opamps, I designed a PCB to use the I/V converter specified by the DAC manufacturer (the Cambridge has dual AD1955 DACs).

I initially used an NE5534 as the I/V converter, but found that the OPA2134 was actually some 3dB+ quieter (despite having an inferior noise figure on paper) and this was attributed to the matched source impedance from the DAC chip combined with the higher input currents on the 5534 (the OPA is a FET opamp hence low input noise currents).

Dicks Website is interesting, although there's no mention of CMRR or the effects of an unbalanced* source impedance on input noise.

• meaning the inputs of the opamp being unbalanced - not a balanced connection.

Interesting...........thanks for that Tony.

That A07 review on ASR is a little odd..........the result was solid/good when powered by the 32V 5A 160W supply. The unit delivered 2 x 77W with decent noise performance etc. The result wasn't great with the relatively weak 48v 3A 144W supply............I'd like to see the A07 measured with a supply that isn't limiting the test results.

Edit. https://www.audiosciencereview.com/forum/index.php...

Edited by Crackie on Tuesday 7th December 11:05

[TonyRPH]

Crackie said:

Interesting...........thanks for that Tony.

That A07 review on ASR is a little odd..........the result was solid/good when powered by the 32V 5A 160W supply. The unit delivered 2 x 77W with decent noise performance etc. The result wasn't great with the relatively weak 48v 3A 144W supply............iirc the TPA325# series chips both have a 3A low current threshold; I'll check that out when I've a little more time. I'd like to see the A07 measured with a supply that isn't limiting the test results.

Amir did make a point about the protection circuit shutting it down preventing him from testing for more power.

My understanding of these desktop amps is that the heatsinks are undersized restricting maximum power output.

I do remember when I was testing the el cheapo TPA3116 boards i bought a while back - the heatsinks got uncomfortably hot with prolonged testing, even at moderate power. I was running them at 25Vdc getting 60w into 4 ohms at the onset of clipping - at which point they were drawing 2.7A (this was in BTL mode).

The 3116 is rated for 26Vdc maximum, so I was probably pushing the boundaries a bit!

[Crackie]

TonyRPH said:

Amir did make a point about the protection circuit shutting it down preventing him from testing for more power.

My understanding of these desktop amps is that the heatsinks are undersized restricting maximum power output.

I do remember when I was testing the el cheapo TPA3116 boards i bought a while back - the heatsinks got uncomfortably hot with prolonged testing, even at moderate power. I was running them at 25Vdc getting 60w into 4 ohms at the onset of clipping - at which point they were drawing 2.7A (this was in BTL mode).

The 3116 is rated for 26Vdc maximum, so I was probably pushing the boundaries a bit!

So many threads on the ASR forum discussing and measuring the A04 / A07. They both appear to be very well regarded........with the caveat that one of the better low noise/low ripple SMPS units, or a competently engineered linear supply, is used to feed them.

Edited by Crackie on Tuesday 7th December 13:14

[Crackie]

TonyRPH said:

I recently modified a Cambridge 851D DAC (not through choice, it had developed a nasty fault, common to the model and it's variants). Rather than spend £££ on the original surface mount opamps, I designed a PCB to use the I/V converter specified by the DAC manufacturer (the Cambridge has dual AD1955 DACs).

I initially used an NE5534 as the I/V converter, but found that the OPA2134 OPA134 was actually some 3dB+ quieter (despite having an inferior noise figure on paper) and this was attributed to the matched source impedance from the DAC chip combined with the higher input currents on the 5534 (the OPA is a FET opamp hence low input noise currents).

Dicks Website is interesting, although there's no mention of CMRR or the effects of an unbalanced* source impedance on input noise.

• meaning the inputs of the opamp being unbalanced - not a balanced connection.

ETA: Should have put OPA134 not 2134! (the 134 is a single opamp - 2134 dual).

Edited by TonyRPH on Tuesday 7th December 10:52

Tony.........re-op amps. Some comprehensive test info. here if you've not seen already. https://nwavguy.blogspot.com/2011/08/op-amp-measur...

[TonyRPH]

Crackie said:

Tony.........re-op amps. Some comprehensive test info. here if you've not seen already. https://nwavguy.blogspot.com/2011/08/op-amp-measur...

Yes, I followed that blog many years ago. Sadly Nwavguy disappeared into the ether.

I have carried out some similar tests myself here OPAMP tests into a 2k7 load and here OPAMP tests 270 ohms