Hi,
I plan to use an RAC20-15DK to power aa X-Altra HPA-1 build in a relatively small case. Would it be fine to put a CLCLC filter on the output of it to filter out the switching noise? Something like this...
Mainly worried about instability of the AC/DC module as some of them can't handle much capacitance on the output.
I plan to use an RAC20-15DK to power aa X-Altra HPA-1 build in a relatively small case. Would it be fine to put a CLCLC filter on the output of it to filter out the switching noise? Something like this...
Mainly worried about instability of the AC/DC module as some of them can't handle much capacitance on the output.
Last edited:
You will be curing a self created situation. Maybe a linear PSU makes things simpler as the part count is more or less the same as the filter you propose.
Can't really fit a transformer in the amp chassis I want to use, only 160x140x45mm outer dimensions (150x130x41mm) for a class A amp and 15VA PSU!
Then your next challenge is where the heat will be dissipated. Essentially the casing is chosen too small so now concessions must be made to PSU choice and heat dissipation....
Which then poses the question what the final outcome will be. In such a small casing a class AB HPA and a PSU of a few VA would be a better choice. There is a chance that the total sum of choices and following decisions will not give the desired outcome. Even the best HPA in the world will not perform optimally with a switcher.
Which then poses the question what the final outcome will be. In such a small casing a class AB HPA and a PSU of a few VA would be a better choice. There is a chance that the total sum of choices and following decisions will not give the desired outcome. Even the best HPA in the world will not perform optimally with a switcher.
Last edited:
The main heat-producing BJTs will be sinked directly to the amp chassis so no problems there (5.2w heat dissipation for both channels as I'm putting an X-Altra in there), the only issue to be solved is PSU, hence let's return to the original question. 😛
Last edited:
What makes you think that? I have seen nothing that would suggest an SMPS would perform worse. Stuff like the Singxer SA-1 uses two Meanwell IRM-20-24 modules in its PSU, for example, and it performs admirably. I simply lack the knowledge to know whether they can deal with an additional CLCLC filter on the output. 😀
Last edited:
Maybe the need for a postfilter to filter out unwanted HF/RF hash (produced by the PSU itself 🙂) makes me think that….
The choice for issues dictates the need to solve those issues.
The choice for issues dictates the need to solve those issues.
Last edited:
Would be CLCLC really necessary? Your amplifier uses composite topology with opamp input which should provide plenty of PSRR.
This statement is false. Switching power supplies are completely adequate for powering even best performing amplifiers and there are lots of designs proving that.
You are right. I forgot to add "compared to using linear power supplies with recent regulators that have below 1 µV noise numbers".
It is for a headphone amplifier in this case, not a power amplifier. SMPS may be "adequate" but linear power supplies will be "more adequate". Any simple comparison will make you say the same.
All hypothetical of course as OP wants to put a class A HPA and 15VA PSU in a 150x130x41mm casing. Now THAT is a challenge.
It is for a headphone amplifier in this case, not a power amplifier. SMPS may be "adequate" but linear power supplies will be "more adequate". Any simple comparison will make you say the same.
All hypothetical of course as OP wants to put a class A HPA and 15VA PSU in a 150x130x41mm casing. Now THAT is a challenge.
Last edited:
That is a small chassis but it's possible especially using 3D wiring (point to point).
I built a phono stage with PSU regulation that uses 10 dual triodes and a DC-DC board in a chassis that is only 127x229x52. If I can do that, OP can build an SS HP amp in a Sucrets tin. 😀 Ok, maybe not THAT small of a tin if they want Class A... 🙂 I did build a 1WPC stereo amp using LM386 in a Sucrets tin powered by a 9V battery once. Pocket headphone amp. Since it was designed for speakers, there was LOTS of hiss though.
I built a phono stage with PSU regulation that uses 10 dual triodes and a DC-DC board in a chassis that is only 127x229x52. If I can do that, OP can build an SS HP amp in a Sucrets tin. 😀 Ok, maybe not THAT small of a tin if they want Class A... 🙂 I did build a 1WPC stereo amp using LM386 in a Sucrets tin powered by a 9V battery once. Pocket headphone amp. Since it was designed for speakers, there was LOTS of hiss though.
Last edited:
The post filter requires only as many parts as is required to filter out noise from the mains line as you previously mentioned 🙂 and as for the 'simple comparison', the benefits are much smaller size and better efficiency, hence much less thermal management and space is needed. Different engineering challenges, different solutions.
CLCLC is probably not necessary (CLC or LC probably is enough, or even just filtering caps), but I'm designing a PCB to fit into the case and there is room for some filtering with the SMPS modules I'm using, hence this question. 😀
For context, this is the case. The TO220 packages of the MJE15032/33 will mount using directly to the case those screws on the side. Heat dissipation should be sufficient, and I double checked with Bonsai just in case too.
Now, can we please move on to the original question?
MJE15032/33 for headphones? What for? I have a 1200W power amplifier that uses these for the outputs (8 per channel, TO-3). It just seems like overkill. Based on the PSU schematic (zener shunt), I thought you'd be using opamps. Then again, I'm a tube guy.
A chassis with tubes has the tubes sticking out. Tubes don't mind heat too much as they are the cause of the heat and thus they ask for a different approach. Both comparisons are off. I take the OP states the PSU to be an issue as the heat will be inside the casing as the power transistors will be bolted to the casing. Ah, just when I posted this post #11 popped up.
That is a small casing indeed and I hope the most essential cover (of course not pictured), the upper one, has ventilation holes. Also with class A perforating the bottom is a wise choice.
Last edited:
Ask Bonsai, he's the one who designed the HPA-1 😀 I was wondering why the MJEs myself but the measurements are good so hey. Did have an idle thought about changing them to 15028/29s though.
Those are specced as TVS diodes to protect the amp, just followed the CUI PSK-20D datasheet when adding those to the circuit. Probably not necessary either but they don't hurt.
The PSU was mainly an issue because of size constraints. The lid does not have perforations stock but I will get an acrylic top and bottom that do have holes in them and will use it topless until they arrive.
Ok, I started my comments based on the term "X-Altra HPA-1" and saw quite large heatsinks in the original design. That is why I think convection is needed. Each channel has an idle current of 90 mA. Using acrylic covers with a hot device again is a design choice contrary to normal engineering.
Back to the original question: I think using 10 µF 16V for C19 and C20 (or omitting C19 and C20) prevents what you are afraid of.
*I recommend to not connect Audio GND to PE directly. It is drawn like that in post #1 or do you like to introduce issues so you have something to solve? 😀
Back to the original question: I think using 10 µF 16V for C19 and C20 (or omitting C19 and C20) prevents what you are afraid of.
*I recommend to not connect Audio GND to PE directly. It is drawn like that in post #1 or do you like to introduce issues so you have something to solve? 😀
Last edited:
Two devices share one heatsink, and as mentioned previously, its only 5.2w heat dissipation for both channels.
I doubt the amp will get so hot that I won't be able to touch it, but if that's the case then I will use a higher temp plastic.
Before designing a filter, why not check out what kind of noise there is on the output?
You can use a spectrum analyzer, or FFT on a digital scope...
> The TO220 packages of the MJE15032/33 will mount using directly to the case those screws on the side.
If it gets too hot you can always add a bit of thermal grease between the top cover (which I assume is 2mm thick aluminium from the photos) and the case so the top cover also acts as a heat sink...
You can use a spectrum analyzer, or FFT on a digital scope...
> The TO220 packages of the MJE15032/33 will mount using directly to the case those screws on the side.
If it gets too hot you can always add a bit of thermal grease between the top cover (which I assume is 2mm thick aluminium from the photos) and the case so the top cover also acts as a heat sink...
Last edited:
Gotcha, might get rid of L3 and L4 again, I'm aware that a CLCLC filter is probably overkill.
That's chassis ground, not PE. Y2 cap between chassis ground and PE is standard for me, don't worry. 🙂
Good point. The RAC10-K units switch at 100kHz, so I assumed I just needed to do a standard low-pass. I have a Rigol DS1054Z I can use to analyse the noise (once I actually buy the units).
Good suggestion, thank you!
Chassis ground is PE isn't it? That is if you mean the casing itself. That should be connected to PE directly and nothing else. Audio GND should definitely not be connected to the casing.
You're right actually, my schematic is wrong; had a cap there that must've gotten yeeted at some point. Here's what I mean:
