DIY Purifi Amp builds

[realdb]

Where

Back & front

View attachment 95559View attachment 95558

Where did you get the case from

 

[Deleted member 65]

Where

Where did you get the case from

It’s a Silent PC case from HDPlex.com

 

[JimB]

Last week, I finally got to build up my first amp in the Ghent cases I've wanted. Sadly, the amp did not impress. In fact, it was bad. Weak (low output), unexpectedly noisy and lacking bass. I tried different input cables, different source outputs, both channels - no change. Must be the power supply, somehow... Afterall, @phoenixdogfan had SMPS problems. So, available weekend time was spent on ultimately unnecessary troubleshooting. Cutting to the chase, all of the RCA->XLR input cables I bought most recently were bad: signal(-) was soldered to pin 1 and they had no connection to pin 3! After reworking the connections, I now have normal operation. I have no critical listening observations to add, other than to say, it no longer sounds like crap!

I did two types of build: vain (because I wanted it my way); and sane (because it's easier and more than adequate). I'll post pictures and few details later.

 

[NTSOS]

Last week, I finally got to build up my first amp in the Ghent cases I've wanted. Sadly, the amp did not impress. In fact, it was bad. Weak (low output), unexpectedly noisy and lacking bass. I tried different input cables, different source outputs, both channels - no change. Must be the power supply, somehow... Afterall, @phoenixdogfan had SMPS problems. So, available weekend time was spent on ultimately unnecessary troubleshooting. Cutting to the chase, all of the RCA->XLR input cables I bought most recently were bad: signal(-) was soldered to pin 1 and they had no connection to pin 3! After reworking the connections, I now have normal operation. I have no critical listening observations to add, other than to say, it no longer sounds like crap!

I did two types of build: vain (because I wanted it my way); and sane (because it's easier and more than adequate). I'll post pictures and few details later.

Yikes, would you care to divulge the cable brand to avoid?

 

[phoenixdogfan]

Last week, I finally got to build up my first amp in the Ghent cases I've wanted. Sadly, the amp did not impress. In fact, it was bad. Weak (low output), unexpectedly noisy and lacking bass. I tried different input cables, different source outputs, both channels - no change. Must be the power supply, somehow... Afterall, @phoenixdogfan had SMPS problems. So, available weekend time was spent on ultimately unnecessary troubleshooting. Cutting to the chase, all of the RCA->XLR input cables I bought most recently were bad: signal(-) was soldered to pin 1 and they had no connection to pin 3! After reworking the connections, I now have normal operation. I have no critical listening observations to add, other than to say, it no longer sounds like crap!

I did two types of build: vain (because I wanted it my way); and sane (because it's easier and more than adequate). I'll post pictures and few details later.

Pls don't tell me you bought it from Ghent!

 

[JimB]

This was a small, slightly custom 'batch' from ~20 months ago. I just hadn't used any of them until late last week. Since I've bought good ones of this exact type from them before, and again since, I'll not say the vendor.

 

[JimB]

My "sane" build:
- Ghent binding post case, with standard gold-plated brass CMC posts (not on their website)
- DELTRON-style output sockets and S1 removed
- 1 ohm, leaded feedback safety resistors added
- Feedback from on the post, rather than the solder pool of the output wires
- 15 AWG bus wire output connections (18||18)
- Arctic MX-4 thermal paste under the SMPS and amp modules
- Fused mains filter, with Y-Caps disconnected from ground
- Star quad mains wiring, 15 AWG (18||18)
- Tightly wound SMPS interconnects, with nAMPON and nFATAL implemented
- Just a touch of NOALOX on the chassis under the ground lug.

Simple and clean. Other angles are like shown for the EVAL1 case on the Ghent website.

 

[JimB]

My "vain" build:
- Ghent binding post case, upgraded to gold-plated, solid-copper CMC posts (not on their website)
- DELTRON-style output sockets and S1 removed
- 1 ohm, SMD feedback safety resistors added
- Feedback from a point midway along post (trapped behind the nut), with soldered safety
- Star-quad 13 AWG output connections (tinned 16||16 BELDEN)
- Arctic MX-4 thermal paste “in” and under the SMPS, as well as under the amp modules
- Added SMPS heatsinking
- Fused mains filter, with Y-Caps disconnected from ground
- Grounded-shield, star quad mains wiring, 15 AWG (18||18)
- Tightly wound SMPS interconnects, with nAMPON and nFATAL implemented
- Just a touch of NOALOX on the chassis under the ground lug.

The actual solder job is a bit better than it looks in this photo!

 

[beren777]

My "sane" build:
- Ghent binding post case, with standard gold-plated brass CMC posts (not on their website)
- DELRON output sockets and S1 removed
- 1 ohm, leaded feedback safety resistors added
- Feedback from on the post, rather than the solder pool of the output wires
- 15 AWG bus wire output connections (18||18)
- Arctic MX-4 thermal paste under the SMPS and amp modules
- Fused mains filter, with X-Caps disconnected from ground
- Star quad mains wiring, 15 AWG (18||18)
- Tightly wound SMPS interconnects, with nAMPON and nFATAL implemented

Did you just not like the Deltron-style speaker connectors?

Any measured thermal performance improvement with the paste and/or heat sinks?

 

[JimB]

Did you just not like the Delron-style speaker connectors?

Any measured thermal performance improvement with the paste and/or heat sinks?

I think the DELTRON-style are very good. If you want to use them, the standard Ghent EVAL1 case, including speakONs, is a good way to go.
http://www.ghentaudio.com/kit/purifi-eval1-se.html

But, if you want binding posts, and you want to use feedback from them, you must remove the 'DELTRONs'. This is what Purifi did on their demo amps. Otherwise, the feedback is from those sockets on the PCB.

The benefit of a little, high quality thermal compound, applied well, will be slightly lower heatsinked component temperatures. How much cooler depends on the exact thermal contact without it, and the power level being used at the time. I didn't go crazy here. Early in the year, I planned to send one of these to Amir to see how whether his power test would run without shutting down (as it did on the Purifi build), but it doesn't seem that interesting after all this time.

I do like the venting details of this Ghent case. It is important for the temperature of the components that are not directly heatsinked, and is very much like what I suggested they use.

 

[JaapDeventer]

Hi, I'm wondering about the 1 Ohm resistors. I have the feedback wires soldered very well. The risk of one getting broken is very low. But what would happen if that one got broken. I read the amp than has an open loop. But would it break the speakers or the amp? Or would it just sound not as good?

Thanks!

 

[beren777]

I think the DELRONs are very good. If you want to use them, the standard Ghent EVAL1 case, including speakONs, is a good way to go.
http://www.ghentaudio.com/kit/purifi-eval1-se.html

But, if you want binding posts, and you want to use feedback from them, you must remove the DELRONs. This is what Purifi did on their demo amps. Otherwise, the feedback is from the DELRONs, on the PCB.

The benefit of a little, high quality thermal compound, applied well, will be slightly lower heatsinked component temperatures. How much cooler depends on the exact thermal contact without it, and the power level being used at the time. I didn't go crazy here. Early in the year, I planned to send one of these to Amir to see how whether his power test would run without shutting down (as it did on the Purifi build), but it doesn't seem that interesting after all this time.

I do like the venting details of this Ghent case. It is important for the temperature of the components that are not directly heatsinked, and is very much like what I suggested they use.

My amp is just an EVAL1 + Hypex SMPS in the standard Ghent EVAL1 case. I haven't soldered the Speakon connectors yet, working my way up to that. A little timid about scorching the buffer board, I'm still a novice with the iron.

I understand the theory of using thermal grease. It'd be interesting to measure it. I think the data sheet says the thermal cutoff is 75C? Do you have a temp probe? Is Amir's power test representative of playing music, even at high levels?

What part did you use for the SMPS heat sink, or did you custom cut one to size?

I'd rather have seen smaller "pinhole" style venting like Ghent uses on their Hypex cases. The longer slot-style venting on the top cover seems like asking for paperclips to fall inside.

 

[beren777]

FYI, here is how Purifi wired their own demo box with dual outputs (on-board DELRONs plus 4-pole speakONs):

View attachment 92604 View attachment 92606
There are two wires for "black", from the two open holes in the FE02 board, connecting to the top and bottom tabs, . If you do this, solder from the top surface of the PCB - that's where the trace is. The same goes for leads to the side tabs from the "red" PCB holes. They used bare bus wire, but note the addition of red insulation to ensure no accidental output short on the wire that crosses behind the lower one from black to the left-hand tab. I'd solder the wires from the PCB first, before case assembly.

I only intend to use 2-pole Speakon cables, no biwiring in my future. Can I just connect the '1' set of tabs without causing problems? What gauge wire to use?

Will feedback sensing be degraded when using the Speakon connectors, because the sensing occurs at the Deltron connector in this configuration? If I primarily use Speakon in the future, should I remove the Deltrons and move feedback sensing to the Speakon connectors?

(edited to correct my lack of t's)

 

[somebodyelse]

Is 'Delron' just a typo that's stuck? Deltron is a reasonably well known manufacturer of similar terminals:
https://www.dem-uk.com/deltron-comp...ctors/4mm_90deg_pcb_mounted_insulated_socket/

 

[JimB]

Hi, I'm wondering about the 1 Ohm resistors. I have the feedback wires soldered very well. The risk of one getting broken is very low. But what would happen if that one got broken. I read the amp than has an open loop. But would it break the speakers or the amp? Or would it just sound not as good?

Thanks!

That would be really bad. Open loop means maximum gain. Output would go to the limit. I'd expect to blow any common tweeter. BUT, I have certainly not tested that! Anyway, you can argue that any of several single component failures could do the same, or maybe worse. Really, in a properly built unit, it should not be a real risk. It is mostly a risk for DIY - getting it wrong.

Losing the external feedback wire connection WITH a 1 ohm bypass in place would likely cause almost no noticeable change - a slight increase in damping factor. I don't have the feedback circuit details to estimate better than that.

 

[JimB]

My amp is just an EVAL1 + Hypex SMPS in the standard Ghent EVAL1 case. I haven't soldered the Speakon connectors yet, working my way up to that. A little timid about scorching the buffer board, I'm still a novice with the iron.

I understand the theory of using thermal grease. It'd be interesting to measure it. I think the data sheet says the thermal cutoff is 75C? Do you have a temp probe? Is Amir's power test representative of playing music, even at high levels?

What part did you use for the SMPS heat sink, or did you custom cut one to size?

I'd rather have seen smaller "pinhole" style venting like Ghent uses on their Hypex cases. The longer slot-style venting on the top cover seems like asking for paperclips to fall inside.

Yes. Soldering those wires onto the top layer of the board, well, with the DELTRON-type sockets in place, is not trivial.

Measuring thermal paste improvement requires two, precisely comparable builds, assemble the two different ways. I'll likely do a simple temp check with a low test load at some time...

I this case (pun intended), I used Assmann WSW V2032N, bonded by industrial thermal epoxy after good surface prep. I don't recommend doing something like this unless you are good at it, and you are looking for improved thermal margin, and/or component life, running sustained high outputs, and/or with the amp in a warm area (like a cabinet - in the summer). This was my VAIN, no limits build.

I argued, strongly, to 'improve' on the tiny holes Ghent used in the past. They are not very conductive. The slots are much, MUCH better, thermally. I think you just need to become a qualified paperclip operator!

 

[JimB]

Is 'Delron' just a typo that's stuck? Deltron is a reasonably well known manufacturer of similar terminals:
https://www.dem-uk.com/deltron-comp...ctors/4mm_90deg_pcb_mounted_insulated_socket/

YES! It (DELTRON) was molded in the tops of early sockets we saw pictured, but not once they went into production (none of my 16 sockets are marked at all). I suspect they are not the brand used anymore. My references should be "DELTRON-style". I'll fix some of those as I see them...

 

[beren777]

Measuring thermal paste improvement requires two, precisely comparable builds, assemble the two different ways. I'll likely do a simple temp check with a low test load at some time...

My unit is pretty much identical to yours, just without paste or additional heat sinks. If you describe your test method (including probe locations) I can try to do the same and measure. There will still be a lot of variables (different test equipment, different ambient temp, etc), but I'm happy to give it a try.

 

[JimB]

My unit is pretty much identical to yours, just without paste or additional heat sinks. If you describe your test method (including probe locations) I can try to do the same and measure. There will still be a lot of variables (different test equipment, different ambient temp, etc), but I'm happy to give it a try.

I'll keep that in mind. The most useful tests will be with top cover in place.

 

[phoenixdogfan]

AMPEN doesn't look right. It looks like the PS is going into standby mode. Need you to measure all these relative to GND and report back :

SMPS Stand-by
Amp Standby
DC-Error
Amp Enable

See page 7 of the Hypex PS datasheet for signal details.

https://www.audiosciencereview.com/forum/index.php?attachments/smps1200axx0_09xx-pdf.49157/

Ok. I've looked at this and retested a couple of ten times.

What I am finding to make a long story short is I'm getting 60vDC off the high voltage pins and 12vDC off the VDR connections. Both of those seem within the correct range based off the Hypex PS datasheet p7. But they immediately fall off until reaching 0 Vdc.

The Vaux pins of both J4 and J5 are consistently measuring at 14v Dc, which does not seem correct b/c the data sheet indicates they should be 22v DC. Again the exact same issue: They exhibit this voltage for a split second when the amp is turned on and then fall rapidly to 0v DC.

Amp Standby on J5 shows 0v DC from the moment power turns on
SMPS Standby on J5also show 0v DC.

Amp enable on J4 also shows zero.

Finally, and this is the joker in the deck, DC Error on J4 flashes 14v DC when switched on before dropping to zero.

Based on the Hypex Data sheet I would not expect to read any voltage from Amp Standby, or Smps Standby because those are both input pins, so that seems normal.

However, Amp enable is an output pin and it appears to be outputting zero voltage when it should be outputting a 50v.

Finally, the DC error pin is an input pin. I am clueless why it outputs a 14v DC surge when the amp is powered up.

It does appear the behavior is consistent with the SMPS going into standby even though it isn't connected to anything. I'm inclined to pitch the thing into file 13 and order a new one.

Anyone have any ideas?