Neurochrome HP-LOAD Review (Headphone Dummy Load)

[tomchr]

Fabulous product Tom.

Thanks John. I appreciate it.

In case someone wonders about the differently looking front plate of the HP load in the above linked video: The unit has a lot of extra space inside and is quite easy to modify. So I [...]

That's pretty neat. And thanks again for your recommendation of the XLR-to-3.5/4.4/6.3 mm adapters.

My experience with the AP remote was less stellar, but that is not caused by the HP Load. The AP software has some unexpected restrictions. For example when doing a THD+N sweep versus level you can not set a THD+N max value where the sweep would stop - for example 1%. You are into clipping then and don't want to see crazy lines going up. Multiple impedance sweeps that have the X axis in Watts are cumbersome to do (that's why you don't see them on ASR). Automation of this all quickly lets you do it manually again to get nice looking graphs.

Yeah. That's the only real annoyance I have with the APx software. If you want one graph that shows THD+N vs output power for multiple load resistances you basically need to record THD+N vs output voltage for the different resistances and massage the data in Excel after.

It's in production testing of headphone amps where the automation feature of the HP-LOAD really shines. I run all my amps through a script where the THD+N vs output power is verified for various load resistances. The script collects several other data points as function of load resistance as well. It's very handy to be able to click a button and come back 15 minutes later to a fully tested amp and a test report with a bunch of green checkmarks. This saves a lot of time and prevents many errors. In the past I used clip leads on a resistor bank and would have the APx software prompt me to swap the leads around. That got old real fast!

Tom

 

[MC_RME]

Yeah. That's the only real annoyance I have with the APx software. If you want one graph that shows THD+N vs output power for multiple load resistances you basically need to record THD+N vs output voltage for the different resistances and massage the data in Excel after.

No need for Excel as final diagram generator. You have to import the exported graphs (of the different Ohm measurements) all back into the final measurement. They will then be correctly showing Watts, due to the nature of the export data content. Try it...

 

[tomchr]

Oh... I will try that. Thanks! I paid a lot to get that AP logo on the graphs so generating graphs with Excel has always rubbed me the wrong way.

Tom

 

[DualTriode]

My preference is to see dBV or perhaps dBr vertical and Volts (not Watts) on the Horizontal axis.

Tom I would like to see a FFT plot for 4 volts into 300 Ohms, you know the dashboard test that @amirm uses.

Thanks DT

 

[tomchr]

Yeah. I need to add measurements to the product page.

Tom

 

[xnor]

amir, have you calibrated/measured the actual load resistances for accurate results, especially for accurate output impedance calculations?
The resistors have 1% tolerance and the wires and adapters add some more resistance.

 

[xnor]

You know full well the dodgy Chinese resistors are absolute garbage for ultra low THD measurements. You also know full well the requirements for HPAs are extreme precision at low powers- something you cannot achieve with the types of loads you are advocating.

The actual resistors used in this box are like ~5 € each on mouser.

 

[tomchr]

That sounds about right. The relays cost about the same and there are 22 of them. The parts cost adds up rather quickly. Having a low quantity of boards assembled by a team of pros wasn't cheap either. Such is life.

The HP-LOAD uses a Kelvin connection to sense the output voltage of the amp under test directly at the amp output. This basically takes the cable resistance out of the equation. One could measure the output impedance of an amp by measuring the difference in output voltage vs frequency at 600 Ω vs 32 Ω on the HP-LOAD. That would eliminate the cable resistance as a confounding variable.

I'm not sure if @amirm uses the HP-LOAD for the output impedance measurements or if he's using his own setup. I do know that his old setup adds some cable resistance to the measurement of the output impedance, which is why you see that measurement bottom out at about 0.1 Ω.

Tom

 

[restorer-john]

That sounds about right. The relays cost about the same and there are 22 of them. The parts cost adds up rather quickly. Having a low quantity of boards assembled by a team of pros wasn't cheap either. Such is life.

The HP-LOAD uses a Kelvin connection to sense the output voltage of the amp under test directly at the amp output. This basically takes the cable resistance out of the equation. One could measure the output impedance of an amp by measuring the difference in output voltage vs frequency at 600 Ω vs 32 Ω on the HP-LOAD. That would eliminate the cable resistance as a confounding variable.

I'm not sure if @amirm uses the HP-LOAD for the output impedance measurements or if he's using his own setup. I do know that his old setup adds some cable resistance to the measurement of the output impedance, which is why you see that measurement bottom out at about 0.1 Ω.

Tom

Well said, and nicely restrained Tom

 

[MC_RME]

No need for Excel as final diagram generator. You have to import the exported graphs (of the different Ohm measurements) all back into the final measurement. They will then be correctly showing Watts, due to the nature of the export data content. Try it...

To be more clear:

- Measure (sweep) the DUT in dBu or V with all load values (impedances). No need to set anything specific. You will have 7 graphs ( from 600/300/50/32/20/16/12 Ohms).

- Change the x axis to unit Watts.

- Set 600 Ohm in the References section, Input References, Analog W(atts). The 'voltage' based curves are now scaled/positioned as if they were measured with a 600 Ohm load, which of course is only true for one of the curves, so only that one is shown correctly.

- Export that curve as cvs. Check the file - it will not have dBu or Volt, but W (Watts) inside.

- Do the same (change ref impedance, export respective graph) with the other measurements down to 12 Ohms, so that you have 7 files. There is no way to do a simple group export. It must be done one after the other.

- Once done set the Ref impedance to a value that makes most sense, like 32 Ohms as you will not be able later to change it anymore.

- Now import all 7 graphs one after the other (no group import possible).

- The plot now shows the correct Watts for all 7 impedances.

- Changing the x axis back to dBu screws it. As does changing the ref impedance, so keep the state of Watts and just save the project.

Here is an example on how it looks like:

Note the (maybe) unexpected shift of the graphs. Other than with (always identical) voltage Watts makes the THD+N Ratio value different per impedance, as the voltage is different. So the graphs all have a vertical offset.

 

[staticV3]

Lol, I'm used to these kinds of shenanigans with REW and other free software.
Didn't think AP would sell you their software for $8K and still make you do this shit.

 

[boXem]

Nice sneak peak @MC_RME

 

[Dan Clark]

Lol, I'm used to these kinds of shenanigans with REW and other free software.
Didn't think AP would sell you their software for $8K and still make you do this shit.

AP software is awesome but there are some functions one would want that they just don't have, such as the ability to make groupings of measurements within a set to create multiple averages, per the prior note you have to create the averages from separate test set files, export then import. It's pretty clunky.

We have a 555B on the way and have already have one of Tom's boxes, I'm looking forward to being able to measure flea farts on jet engines, as Mr. Moffat once said.

 

[tomchr]

I also have only praise for the AP software (which is now $3500, not $8k) and the AP hardware (especially the 555B). It's very rare that I enjoy using software, but the APx500 software is a true joy to use. I've used the previous SYS-2700 software which is awful.

Thanks @MC_RME for the procedure. I'll give it a whirl. I think I'll find it handy for speaker amps. The vertical offset of the curves is to be expected if your amp's performance is dominated by the +N and not by the THD. As you point out, 10 mW into 600 Ω requires much higher voltage than 10 mW into 12 Ω, so the SNR is much higher for the 600 Ω measurement and the THD+N lower as result. If the amp's performance was dominated by the THD (and not the +N) you'd probably have the curves overlapping.

One can then contemplate whether showing above plot or one that shows THD+N vs output voltage is the better way to present the data. You could have confused customers either way.

Tom

 

[MC_RME]

Once you go through all the above process you end up with a project having tons of 'Measurement' entries (graphs) and 'Imported' ones. All you need to remember is the Ohm value with which you imported them (I chose 32 Ohms as it is easy to remember, and this most common value is not required then to be exported/imported at all). Now if you want to see the voltage based graphs only make the 'Measurements' visible, and if you want to see Watts only select the 'Imported' ones (while 32 Ohms is the reference. For readers who don't know the software: this is a label in the graph list, so easy to see what is what).

It's very cumbersome to get there, but once there it starts to become fun as you can easily show tons of nice stuff, do comparisons and configure the visual output as much as you like.

Furthermore you can also continue to add measurements to the project, if you just remember to always set it to 32 Ohms (in my case) when importing Watt graphs. What I also really like is that you can load the project into the (free) software running in demo mode (no dongle or APx required), then fiddle with whatever selection and scaling and colours etc, then save the measurement pic. I often do this kind of offline work to produce plots as needed.

 

[lashto]

Now, if any of you would like to buy my APx525, I'm happy to sell ... in 4-5 weeks. I figure I'd get ~$10k on the used market.

nice LOAD there, many thanks and congratulations.
Hope you get lucky with your 525 https://www.ebay.com/sch/i.html?_from=R40&_nkw=apx525

 

[tomchr]

Thanks.

The APx525 sold over a year ago.

Tom

 

[DualTriode]

AP software is awesome but there are some functions one would want that they just don't have, such as the ability to make groupings of measurements within a set to create multiple averages, per the prior note you have to create the averages from separate test set files, export then import. It's pretty clunky.

We have a 555B on the way and have already have one of Tom's boxes, I'm looking forward to being able to measure flea farts on jet engines, as Mr. Moffat once said.

I also have only praise for the AP software (which is now $3500, not $8k) and the AP hardware (especially the 555B). It's very rare that I enjoy using software, but the APx500 software is a true joy to use. I've used the previous SYS-2700 software which is awful.

Thanks @MC_RME for the procedure. I'll give it a whirl. I think I'll find it handy for speaker amps. The vertical offset of the curves is to be expected if your amp's performance is dominated by the +N and not by the THD. As you point out, 10 mW into 600 Ω requires much higher voltage than 10 mW into 12 Ω, so the SNR is much higher for the 600 Ω measurement and the THD+N lower as result. If the amp's performance was dominated by the THD (and not the +N) you'd probably have the curves overlapping.

One can then contemplate whether showing above plot or one that shows THD+N vs output voltage is the better way to present the data. You could have confused customers either way.

Tom

Hello,

For grins I will reply to this APx stuff.

I am not thrilled with AP In November of 2018 I purchased a APx555. I asked AP if a new audio analyzer was going to be released. The answer was no, a month later the 555B was announced.
All about the same time AP changed ownership and the APx500 software became a subscription. Any new functions are dribbled out, each with a version number change and thousands of dollars upcharge.

I call AP GRAS for a microphone or a APx1701 or a AUX 0040 the Northern California sales rep does not return my phone call. At this point I call the main office in Oregon to get a phone call from the NorCal sales Rep. I finally placed an order for the AUX 0040, as in paid the invoice, and they still lost my order. I am not a large commercial company. I still expect them to return a telephone call but that remains difficult.

I have a APx555, 45CA headphone test Jig, an assortment of GRAS microphones and amplifiers, APx1701 and a AUX 0040 all for my small hobby lab.

Keysight does a much better job.

Thanks DT

 

[amirm]

My favorite AP salesperson left and no one contacted me with the replacement. The subscription is horrible with minor features promoted as full releases. Definitely not happy with them.

 

[MC_RME]

The 'improvements' in the full releases seem to never be the stuff that you really need and want...not to mention that they still (version 8.0 now released) don't measure THD as anyone else does.

@DualTriode > Keysight does a much better job.

I wouldn't touch Keysight with a barge pole:

Keysight officially lost the plot - don't buy if you're a hobbyist - Page 1

Keysight officially lost the plot - don't buy if you're a hobbyist - Page 1

www.eevblog.com

Matching to that their overpriced handheld multimeter was a complete failure (IMHO):

As so often it is not easy to find something that just does everything right.