Hi All
I'm looking for some low noise amplifiers to use on some horns. I am currently using a Rotel RB976 which is surprisingly quiet but I'm looking to try out some low power amps and see a lot of options out there from chip amps to low power Class D modules.
Impedances are 8/16/16ohm and I would like to try out some amps starting from 5-10W and up. The sensitivities are from 109dB for the tweeter to well over 110dB for the mid-bass.
What options are very quiet when matched with a fair transformer supply or are SMPS options quieter?
I appreciate your input.
Martin
I'm looking for some low noise amplifiers to use on some horns. I am currently using a Rotel RB976 which is surprisingly quiet but I'm looking to try out some low power amps and see a lot of options out there from chip amps to low power Class D modules.
Impedances are 8/16/16ohm and I would like to try out some amps starting from 5-10W and up. The sensitivities are from 109dB for the tweeter to well over 110dB for the mid-bass.
What options are very quiet when matched with a fair transformer supply or are SMPS options quieter?
I appreciate your input.
Martin
TPA3116 boards I've tried are quiet... Though subject to a turn on thump!
WLX board or the "new" "twin chip" board are good. Run off a 16V ex-Thinkpad SMPS.
WLX board has a mute switch, the twin/dual has a combined on/off/volume pot. Either suffice to alleviate the thump.... Though you have to remember to mute the WLX!
WLX board or the "new" "twin chip" board are good. Run off a 16V ex-Thinkpad SMPS.
WLX board has a mute switch, the twin/dual has a combined on/off/volume pot. Either suffice to alleviate the thump.... Though you have to remember to mute the WLX!
Thank you. How loud is this turn on pop and is there an example circuit to manage the turn on pop circuit externally?
How do these boards fare with quieter PSUs or against a generic chip-amp such as a LM1875/3886 other than the reduces PSU requirements?
How do these boards fare with quieter PSUs or against a generic chip-amp such as a LM1875/3886 other than the reduces PSU requirements?
Then there are the compound amps like the Modulus 86 which a friend has and I want to try. Freedom from buzz is a major thing with such horns as most amps have an audible buzz in their risidual with some horns (if it's quiet enough it doesn't matter once the music is playing but many people expect dead quiet from a system these days)
IcePower50ASX2 SE maybe an option ?
Those modules are very quiet and produce no turn on thump.
FIY I don't own horns but conventional speakers.
http://www.diyaudio.com/forums/class-d/190311-icepower-50asx2-modules.html
ICEpower ? Class D audio amplifiers ? Consumer & Professional Audio
Those modules are very quiet and produce no turn on thump.
FIY I don't own horns but conventional speakers.
http://www.diyaudio.com/forums/class-d/190311-icepower-50asx2-modules.html
ICEpower ? Class D audio amplifiers ? Consumer & Professional Audio
I have my big old,efficient,Tannoy Corner Yorks connected to one of these amps.
1pc TDA7297 Version B 2*15W Digital Audio Amplifier Board Dual Channel AC/DC 12V-in Integrated Circuits from Electronic Components & Supplies on Aliexpress.com | Alibaba Group
I connected the amp to my Tannoys as an experiment 2 years ago and there it has remained, i just love it. The amp is absolutely dead silent, in fact, I leave it on all the time so no worry about switch on thumps.
I can stick my ear right up against the grill and hear nothing. i'm using an inexpensive 12v brick power supply.
With that kind of sensitivity, you're asking for low noise.
My suggestions: (will require notable development, however)
Composite using any of the various high-performance precision opamps (I'd lean towards a FET-input) and any of the various DSL line drivers (run a couple in parallel).
Or, given the power needs, a +/- 18V opamp driving a discrete OPS (the newly announced TI-former BB chip might work well here).
My suggestions: (will require notable development, however)
Composite using any of the various high-performance precision opamps (I'd lean towards a FET-input) and any of the various DSL line drivers (run a couple in parallel).
Or, given the power needs, a +/- 18V opamp driving a discrete OPS (the newly announced TI-former BB chip might work well here).
I estimate 109dB sensitivity of the mid-woofer peaking around 114 dB (thankfully it filters itself acoustically above 500Hz depending on the phase plug arrangement), the mid compression driver is rated higher but on a 300Hz mid horn is perhaps around mid 110s mid band and thus the most sensitive. The tweeter is thankfully "only" around 109dB sensitive on its horn with a taper towards HF that I EQ. I guess if I were to measure at 10m/28.3V I'd end up with a different picture but by then I'd have issues with the neighbourhood. Suffice to say, they are exceptionally sensitive, much, much more so than my 15" Tannoys.
I'm going to have a multi-channel attenuator before the amps to get better gain structure as I am multi-amping (Good Olde DCX2496 for now, looking at other options later). It's quite shocking what my old Rotel can get these things to do and easily ask of just about any subwoofer to clip.
I'm going to have a multi-channel attenuator before the amps to get better gain structure as I am multi-amping (Good Olde DCX2496 for now, looking at other options later). It's quite shocking what my old Rotel can get these things to do and easily ask of just about any subwoofer to clip.
Thank you for all the responses. B&O rate the noise floor at 20 uV A-weighted which is numerically lower than the UCD-180 at 45 (unweighted) or the small 32W UCDs at 30 uV (unweighted). At a guess these are sufficiently quiet for me, though one may still perceive a tiny residual when up close (buzz is more noticeable than thermal noise).
At a guess the PSRR may factor in substantially as well the the ability to properly integrate these circuits which may be a RF thing as these things are probably measured on lab supplies under perfect conditions. I could always look at winding some chokes for a conventional supply, though I guess there may be limitations to that approach and the DCR of a choke may not be wanted?
Last edited:
The TI 3116D2 sheet states a noise floor around 65 uV, the TPA 3251D2 likewise.
ST don't spec the TDA7297 unfortunately.
The Modulus 86 is specified as hum at -105 dBV and noise at -135dBV (I guess not integrated, or?) which works out to 5.6 uV. Unfortunately I am not qualified to develop a composite amp, though I enjoyed reading how even PCB layout could effect orders of magnitude improvements to a LM3886.
Realistically, if I assume noise at 1m distance as my target, then 115dB @ 2.83V is a pretty tough taskmaster with -100dB noise levels re 1W resulting in 15dB which is fine. Even a 20dB or 30dB noise level would still be acceptable to me if it was noise, though if the amp had 13-16dB headroom above 1W then the noise floor starts to creep up to where it starts to intrude without signal applied. Buzz is more objectionable to me so one may have to filter that out more in the PSU. Lots of C, SMPS or CLC?
ST don't spec the TDA7297 unfortunately.
The Modulus 86 is specified as hum at -105 dBV and noise at -135dBV (I guess not integrated, or?) which works out to 5.6 uV. Unfortunately I am not qualified to develop a composite amp, though I enjoyed reading how even PCB layout could effect orders of magnitude improvements to a LM3886.
Realistically, if I assume noise at 1m distance as my target, then 115dB @ 2.83V is a pretty tough taskmaster with -100dB noise levels re 1W resulting in 15dB which is fine. Even a 20dB or 30dB noise level would still be acceptable to me if it was noise, though if the amp had 13-16dB headroom above 1W then the noise floor starts to creep up to where it starts to intrude without signal applied. Buzz is more objectionable to me so one may have to filter that out more in the PSU. Lots of C, SMPS or CLC?
SMPS+CLC; shift supply noise above baseband and filter it. The 60+dB reduction in ripple from switching 1000+ times faster doesn't hurt either.
Your amplifier requirements describe the build I have in progress. The Modulus is going to be quieter than most class D amplifiers but ultimately lower noise requires lower amplifier gain and optimizations for lower signal swing. This favours class AB, particularly with 110 dB SPL/W drivers. However, what's the output swing of your preamp or DAC?
Your amplifier requirements describe the build I have in progress. The Modulus is going to be quieter than most class D amplifiers but ultimately lower noise requires lower amplifier gain and optimizations for lower signal swing. This favours class AB, particularly with 110 dB SPL/W drivers. However, what's the output swing of your preamp or DAC?
Yep, not integrated; scroll to the noise floor figure about 80% of the way down the page here. Integrated should be around 25uV, though I need to check my maths.
BJT based parts tend to offer lower noise in these kinds of applications, actually; the lower signal swings allow feedback impedances below the BJT-FET transition point. The drive levels here are high enough thanks to the 16 ohm drivers there's probably not too much difference, though. I'd have to run the maths.
Last edited:
I should have mentioned that, yes, the low feedback impedance probably means a BJT-input opamp is a better choice (from a noise-perspective) for this location. I'd prefer a dual FET-input opamp in this place because I can use the second channel as my DC servo, keeping things more compact (parsimony!).
So I agree, and also agree that it probably wouldn't be too much different noise-wise (and much quieter than a monolithic power-chip part).
Turn on pop is a full blooded thump.... There has been mention of circuits in the TPA3116D2 topic but I haven't seen any details.
My setup are essentially silent with no load, no signs of hiss etc. Though I'm running Tannoy E11s, so much less sensitive than your speakers.
J.
Turns out I was a bit low. Tom just let me know he's added an integrated spec to the Modulus 2.0 page. Higher than nCore but the Modulus is more than competitive on other parameters. Not sure if anyone's published a noise measurement on a Liteamp but the sims point earlier in the thread suggest around 50 uV.
Other low noise class D amps I'm aware of are the TAS2555 (16 uV), SSM2315 (21 uV A), LM4675 (22uV A), and MAX98357 (25 uV A). These don't meet the output requirements here, though.
It's an approach I like as well. In the sims I've run I've found if the servo's optimized for BJT devices the performance is comparable to a FET servo. Maybe a little bit higher residual offset but the noise difference is negligible due to the low corner frequency of the servo filter.
If typical offsets of a few hundred uV are acceptable---which follows more or less by definition for these sorts of designs---one doesn't necessarily need to use a servoed topology. With some careful design one can take advantage of this to reduce the amp's output noise a bit more. Subject to diminishing returns, though, as it becomes increasingly likely the amp will be quieter than the source. Makes designing a DAC output buffer which can take advantage of the amp nontrivial.
- Status
- Not open for further replies.