DIY Purifi Amp builds

[AndyLu]

I have an NC-1200 based amp here (Merrill Audio Veritas twins) it runs very hot in comparison to the Purifi on the same load. The Purfif amps with a single SMPS 1200 run barely warm, even cranking Tool at "11" for an hour on my 4 ohm speakers. Hence I had total confidence in putting my stereo build in a non-ventilated chassis. This is really a night and day difference in terms of heat in the real world. I will try and take a pic of my build later today to share with folks here.

I can be mistaken, but doesn't the Merrill Veritas have a linear power supply instead of a switching one?.

 

[barrows]

I can be mistaken, but doesn't the Merrill Veritas have a linear power supply instead of a switching one?.

No, two NC-1200 modules and two SMPS-1200 in a single chassis in the "Twins" versions.

 

[AndyLu]

No, two NC-1200 modules and two SMPS-1200 in a single chassis in the "Twins" versions.

Ah thanks. I was probably confusing it with another Merrill model then

 

[Deleted member 65]

Planning my Purifi build, I wont be using thermal paste.

Had to dig out some research papers from my big blue days. This is what they said at the time:

When you spread thermal paste across a CPU you lose about 40% of the CPU’s cooling efficiency to the ceramic or metal particles that make up the thermal compound. The CPU’s thermal budget is further depleted by the peculiarity of the paste’s flow behavior. As a CPU heats up, paste particles bunch together along the diagonals of the chip, forming an X pattern of sorts. Obviously, this runs counter to the spread evenly recommendation given for thermal-paste application.

If you decide to use thermal paste make sure you use as little as possible.

 

[barrows]

Here's a pic of my build. I still need to hook up my standby switch, but I have been enjoying listening to it too much to bother! Definitely better than my NC-400 build in sound quality, not huge difference though, but a meaningful one.

I have a custom linear power supply providing the +/- 18 VDC rails, with Belleson discrete regulators. WBT 0708 binding posts, and Neutrik XLR jacks wired with pin 1 direct to chassis. No ventilation and no thermal compound and no problems, it barely gets warm even when blasting at 92 dB avg for an hour. I would not bother with thermal compound at all.

 

[boXem]

Here's a pic of my build. I still need to hook up my standby switch, but I have been enjoying listening to it too much to bother! Definitely better than my NC-400 build in sound quality, not huge difference though, but a meaningful one.

View attachment 63274
I have a custom linear power supply providing the +/- 18 VDC rails, with Belleson discrete regulators. WBT 0708 binding posts, and Neutrik XLR jacks wired with pin 1 direct to chassis. No ventilation and no thermal compound and no problems, it barely gets warm even when blasting at 92 dB avg for an hour. I would not bother with thermal compound at all.

Congrats for the build, very nice.

With 80 dB/W speakers, 92 dB is 4 W, for sure it stays cool .
For comparison, 120W continuous would be 121 dB with the same speakers. Completely dumb for home environment, but if your cooling works with this, you are covered for everything.
Concerning thermal compound, I agree that too much is worse than none.

 

[mocenigo]

Yes, pin 1 will be at ground, just make sure that the heat sinks under the modules make good contact with the chassis, you should remove some paint/anodizing on the chassis to be sure. It is still probably better to connect pin 1 directly to the chassis, I did this on my build, but I removed the XLR jacks on the eval board, and used separate jacks with wiring.

The electrical contact will be through the screws as well. In my case the 1ET400A has contact to a painted surface (I painted the suspended internal bottom, which I guess will not damage heat transfer too much) but electrically there are these m3 screws and tightening them has removed the paint as well, so electrical contact is guaranteed in any case. I checked, and the resistance between ground on the Purifi modules and an exposed part of the suspended plate is zero.

 

[mocenigo]

Here's a pic of my build. I still need to hook up my standby switch, but I have been enjoying listening to it too much to bother! Definitely better than my NC-400 build in sound quality, not huge difference though, but a meaningful one.

View attachment 63274
I have a custom linear power supply providing the +/- 18 VDC rails, with Belleson discrete regulators. WBT 0708 binding posts, and Neutrik XLR jacks wired with pin 1 direct to chassis. No ventilation and no thermal compound and no problems, it barely gets warm even when blasting at 92 dB avg for an hour. I would not bother with thermal compound at all.

Wow this is clean. You make my build look messy!

 

[barrows]

The electrical contact will be through the screws as well. In my case the 1ET400A has contact to a painted surface (I painted the suspended internal bottom, which I guess will not damage heat transfer too much) but electrically there are these m3 screws and tightening them has removed the paint as well, so electrical contact is guaranteed in any case. I checked, and the resistance between ground on the Purifi modules and an exposed part of the suspended plate is zero.

Yeah, I just do not like to rely on that small a contact area, so made sure to remove a small amount of anodizing around one screw hole for each module, just to be certain of good low impedance contact. This is important for RF as well, so it is a good idea just to make sure there is good contact. Of course there is still no need to remove tons of paint/anodizing, just a small area around a single screw hole. At the same time, I made sure to remove anodizing on the chassis panels such that every panel is continuous with ground, this is a good idea for safety as well as RF/shielding. The Dremel is your friend here!

 

[JimB]

Planning my Purifi build, I wont be using thermal paste.

Had to dig out some research papers from my big blue days. This is what they said at the time:

When you spread thermal paste across a CPU you lose about 40% of the CPU’s cooling efficiency to the ceramic or metal particles that make up the thermal compound. The CPU’s thermal budget is further depleted by the peculiarity of the paste’s flow behavior. As a CPU heats up, paste particles bunch together along the diagonals of the chip, forming an X pattern of sorts. Obviously, this runs counter to the spread evenly recommendation given for thermal-paste application.

If you decide to use thermal paste make sure you use as little as possible.

Hi, @LarsS! Can you detail more on this? The efficacy of thermal interface transfer 'fluids' is WELL established, as are the basic rules for best utilization. The idea that using it results in a loss of cooling efficiency is strange.

 

[Count Arthur]

The heatsinks need to make electrical contact with the chassis (for grounding and RF reasons), hence the need to remove some of the paint/anodizing.

They will, via one of the screws, if you remove the anidodising around the mounting hole on the outside of the case.

 

[JimB]

Just got a set of screws from Amazon along with allen wrench. Thermal paste under both the heat sinks of the two amps and the Smps is a great idea. Still waiting for the SMPS which is out of stock. When it comes I will have all the pieces.

I am also following the "pin 1" discussion. Does the Eval 1 as used right out of the box properly ground out pin 1? In other words, if I just connect an XLR cable to the jacks supplied with the board, will Pin 1 be properly grounded assuming the board is fastened properly to the bottom of a metal chasis like the Ghent, providing I have a 5 hole back panel fabricated? Or if I just use three XLR pins into each eval 1 jack and then soldering those pins directly to the suplied back panel of the Ghent XLR inputs would that properly ground out the XLR inputs as well, or do I need to do something else?

On my EVAL1, the 'plastic' XLR socket body has a metal rib inside the lower screw hole that is the connector shell contact (what your XLR plug body contacts). This is soldered directly to the EVAL1 "ground plane". Adjacent to this, Pin 1 is soldered to the same copper layer. If you use a metal mounting screw that conducts to the back panel, it will be tied to Pin 1 and the XLR plug body. The metal stand-offs at the outside corners of the EVAL1 are screwed to pads connecting to the same copper plane, so making a conductive connection from the post to the chassis bottom will tie that in, as well.

 

[jae]

How do the mains-powered hypex modules compare to the nc400 and the Purifi? Just more distortion?

 

[phoenixdogfan]

On my EVAL1, the 'plastic' XLR socket body has a metal rib inside the lower screw hole that is the connector shell contact (what your XLR plug body contacts). This is soldered directly to the EVAL1 "ground plane". Adjacent to this, Pin 1 is soldered to the same copper layer. If you use a metal mounting screw that conducts to the back panel, it will be tied to Pin 1 and the XLR plug body. The metal stand-offs at the outside corners of the EVAL1 are screwed to pads connecting to the same copper plane, so making a conductive connection from the post to the chassis bottom will tie that in, as well.

Thanks. So I have at least two ways to ground out pin 1 just by either screwing in the Eval 1 XLX sockets into bare metal on the back panel (assuming I have one fabricated to accommodate the Eval 1), or just by screwing the front stand offs into bare metal on the chasis bottom. Sounds clear enough.

 

[JimB]

Thanks. So I have at least two ways to ground out pin 1 just by either screwing in the Eval 1 XLX sockets into bare metal on the back panel (assuming I have one fabricated to accommodate the Eval 1), or just by screwing the front stand offs into bare metal on the chasis bottom. Sounds clear enough.

Pin 1 is already "grounded". My comments indicate two ways that ground can be tied to the amp case by assembly without separate, special means.

 

[Deleted member 65]

Hi, @LarsS! Can you detail more on this? The efficacy of thermal interface transfer 'fluids' is WELL established, as are the basic rules for best utilization. The idea that using it results in a loss of cooling efficiency is strange.

Done correct in the right application of course you should use thermal compound eg cooling of CPU.

Purpose of thermal compound is to remedy the imperfections on the metal surfaces i.e fill the microscopic air gaps between the two metal surfaces. The thermal conductivity measured in Watts/meter Kelvin is 0.024 W/mK for air, aluminium 205 W/mK & thermal compound varies between ~5-9 W/mK. Replacing air with a thermal compound is a good idea, not so much if you´re replacing aluminium to aluminium contact with a thermal compound.

If correctly applied eg between the Purifi aluminium plate and chassies only filling the microscopic air gaps and not the metal to metal contact then it's a good idea. I don't have the skills to do that ... ;-)

 

[JimB]

Done correct in the right application of course you should use thermal compound eg cooling of CPU. ... If correctly applied eg between the Purifi aluminium plate and chassies only filling the microscopic air gaps and not the metal to metal contact then it's a good idea. I don't have the skills to do that ... ;-)

Thanks for the comments. I want to run a simple comparison of device temperature, at power, without, and then with, Arctic MX-4. Your prediction is that using the MX-4 will raise the component temperature. I doubt that will be the result, but if so, I'll stop using it (saving myself a bit of time and money in the future)! But, don't anyone hold their breath - it will be a couple months before I can do this.

 

[Deleted member 65]

Thanks for the comments. I want to run a simple comparison of device temperature, at power, without, and then with, Arctic MX-4. Your prediction is that using the MX-4 will raise the component temperature. I doubt that will be the result, but if so, I'll stop using it (saving myself a bit of time and money in the future)! But, don't anyone hold their breath - it will be a couple months before I can do this.

You’re most likely much more skilled then me, so my assumption is you’ll succeed in lowering temp ... ;-)

 

[JimB]

You’re most likely much more skilled then me, so my assumption is you’ll succeed in lowering temp ... ;-)

If I "succeed", I'll definitely show how (it won't be sophisticated!).

 

[Count Arthur]

You don't need to be particularly skilled to apply thermal paste with CPUs:

As long as you don't use too little, it seems you can be pretty slap-dash without any detrimental effect.