Beware that the rail splitter needs to provide the full output current. Many of them can't. You could use an opamp with high output current capability driven by a resistive divider from the supply rails as the rail splitter.
Personally I think I'd just use coupling capacitors on the input and output and bias the headphone driver for mid-rail DC on its output. The only drawback is that the output capacitor needs to be pretty substantial. 2500 uF if you want a 2 Hz lower cutoff with 32 Ω headphones. At least a 6.3 or 10 V type would work and could be somewhat reasonable in physical size.
Tom
Entirely to be expected given the DS specs for the TDA. You don't need that 39dB of gain in this application, given the output of the AD9833 is about 200mV and the HPs need about 1V max. Hence 14dB gain is quite sufficient but even with an optimal gain structure, the bottleneck is the -60dB quantization noise of the AD9833.
Sure, I'll upload a photo on Monday, I don't have it here with me
Sure, I understand that the gain is a lot.
but first I would like to understand and solve the problem of the noise in the amplifier, before continuing with the rest.
and that the amplifier reaches 110db spl @32R
I need to connect the headphones to the amplifier circuit and not hear anything when it is in standby.
in all three circuits, with different ICs (TDA2822, TDA7052, PAM8403), I hear the same noise, like when a power amp is at maximum volume and you hear that noise like a wind.
in my tests I have only tested the amplifier circuit, without connecting the AD9833.
maybe I'm missing something. but I always build the circuit according to the data sheet of each one.
What would be the common opamps that you say?
I'm going to order at Mouser, and I can order any
First off you'll need an amp that has less than the normal amount of gain (26dB would be typical), you want 14dB. This precludes use of pretty much any chipamp as their gain (and by extension, their noise) is too high. There are some chipamps which are configurable via software for lower gain but I would guess that complexity is something you'd wish to avoid.
So you're left with building an amp from chips which are going to support lower gain - the opamps which have already been mentioned (OPA1622, OPA1656) will work fine at 14dB as will TPA6120 and AD815 (which hasn't been mentioned). Or if you prefer to stick with a 'ready-rolled' chipamp with too high gain, you could use an output transformer to reduce the gain and the noise.
So you're left with building an amp from chips which are going to support lower gain - the opamps which have already been mentioned (OPA1622, OPA1656) will work fine at 14dB as will TPA6120 and AD815 (which hasn't been mentioned). Or if you prefer to stick with a 'ready-rolled' chipamp with too high gain, you could use an output transformer to reduce the gain and the noise.
perfect and clear
I can handle SPI, I2C, no problem. Could you tell me an IC model that has this feature?
Since I have not defined the amplifier that I am going to use, I have no problem adapting the design to what is best.
my only requirement is to be able to reproduce sine wave from 10 to 110db SPL in my 32R headphones.
Everything else I can change or adapt without problems
The chipamp I had in mind is TDA8594 but it has some other aspects that might well be unsuitable in your application. For example its a quad channel device and its BTL. The quiescent current is high but with software you may be able to turn off unwanted channels. However if you're only using single-ended output that will be a waste of the complementary amp. Its low gain setting is 16dB which probably is close enough to what you want, unfortunately that's for BTL, in SE you'd have 10dB which might turn out to be too low. It still has much more voltage swing than you need - on its 8V minimum supply it easily delivers out 2VRMS.
it's a good chipamp, but size is an issue. and it is true that it is a waste that it has 4 channels, when I only need one.
In practice, I only need one channel working,
only the left side is tested and then the right.
I never need both sides at the same time.
I'm rather suspicious of TI's claims on headphone amp chips after going down the rabbit hole with TPA6120. 98dB SNR doesn't look too promising at first glance, needs a very careful examination of the DS. I'm also dissuaded by the space saving packages which generally aren't DIY-friendly.
MA12070 might be worth a look - although you'll have wasted channels its a real winner in the efficiency stakes. Might be too high output noise though without an OPT and its also not very DIY-friendly to solder.
As you only need the one channel probably OPA1622 is your best bet, available already plonked onto an adapter : https://www.aliexpress.com/item/1005003188313076.html?spm=a2g0o.productlist.main.1.696cCp4TCp4Tgg&algo_pvid=b44e60e3-5c9a-41c9-996f-4dc610feeecc&algo_exp_id=b44e60e3-5c9a-41c9-996f-4dc610feeecc-0&pdp_ext_f={"sku_id":"12000024570663835"}&pdp_npi=3@dis!USD!7.37!6.49!!!!!@21224e9b16779864412656812d074d!12000024570663835!sea!HK!2609685283&curPageLogUid=JsvZSDPTQKmp Worth bearing in mind the handling of the extra pins might not be ideal on an 8pin adapter though.
MA12070 might be worth a look - although you'll have wasted channels its a real winner in the efficiency stakes. Might be too high output noise though without an OPT and its also not very DIY-friendly to solder.
As you only need the one channel probably OPA1622 is your best bet, available already plonked onto an adapter : https://www.aliexpress.com/item/1005003188313076.html?spm=a2g0o.productlist.main.1.696cCp4TCp4Tgg&algo_pvid=b44e60e3-5c9a-41c9-996f-4dc610feeecc&algo_exp_id=b44e60e3-5c9a-41c9-996f-4dc610feeecc-0&pdp_ext_f={"sku_id":"12000024570663835"}&pdp_npi=3@dis!USD!7.37!6.49!!!!!@21224e9b16779864412656812d074d!12000024570663835!sea!HK!2609685283&curPageLogUid=JsvZSDPTQKmp Worth bearing in mind the handling of the extra pins might not be ideal on an 8pin adapter though.
For hearing-loss? Audiometers "usually" generate tone at high level, then a calibrated passive pad to reduce to headphone level. The reasons are also historical. But you should not need and do not want ANY gain to measure hearing thresholds.
TDA2822 (and similar) run about 2uV and gain about 40dB or 1:100, making 200uV hiss at the output.
1mW in 32r= 0.18V=102dB SPL
200uV hiss
I get output hiss OTOO 43dB SPL?? (Somebody check that please.)
An undamaged young ear will hear 43dB SPL fairly easily even without a quiet room. (I'm sitting here about 30dB SPL but it is after midnight a mile out in the woods and there is a foot of snow all around which makes a real difference.) In many typical screening situation the ambient noise (traffic, blowers, etc) will exceed 43dB, perhaps by a lot. I don't think I hear 43dB SPL above 2kHz any more; my threshold rises past 50dB SPL.
If you are screening for hearing loss, 43dB SPL self-hiss is not a serious defect. (Audiologists don't reference SPL but the older HL scale.)
I was thinking just to knock up a super quick breadboard version or quick prototype using 'anything'... LM4562, even a TL071/2 Just try it and get a reference of how quiet you can can get with the phones connected. There is no point getting special devices and making a board for them if you are not going to achieve what you want.
You can even use point to point with a DIP package chip.
is a good starting point.
I'm going to look in my drawers, what IC do I have to do these tests
I don't know which op amp you have, an example of DIP type that is easy to test is
Do you have OPA2134?
It has noise characteristics relatively close to those of the candidate OPA1688.
If the TL072, which is noisier than them, is fine, then the OPA1688 should be fine.
Do you have OPA2134?
It has noise characteristics relatively close to those of the candidate OPA1688.
If the TL072, which is noisier than them, is fine, then the OPA1688 should be fine.
thanks !
Yes. Now I'm waiting for an order from Mouser for several chipamps to do more tests. with all the suggestions they made me.
I will post the progress in these days.