The plan is to use this thread to document my integration of a Zenductor 2 stereo pair into a Modushop Galaxy 1GX388 2U chassis.
Except for removal of the tall Mosfet heatsinks, relocation of the transistors to the chassis sidewalls and reuse of the speaker posts I tried to leave the Zenductor 2s unmodified, and make the assembly as modular as possible. DC power and RCA input connections are via connectorized jumper cables. A preamp mounted to the front panel can be optionally looped into the signal path via rear panel RCA input and output connectors. I integrated two digital bias current displays and a large illuminated push button power switch into the front panel.
Heatsinks and transistors after removal:
They look a bit sad, with their legs cut off. Will they ever run again? ;-)
The amps just fit with the jumper cable connections:
Transistors mounted on sidewalls (1.4 K/W heatsinks, tested in https://www.diyaudio.com/community/...galaxy-1gx388-2u-chassis.426972/#post-7997356):
A quick test to see whether the Mosfets survived:
The 3-digit display shows the bias current. All four transistors survived, and get warm, but not hot. Definitely an improvement over the original heatsinks.
A thermal image:
Lots of filing for the bias display cutouts:
It's been a while since I had to do 'precision filing'; maybe I'll fill the gap later.
Almost finished (except for the preamp connections):
With top cover:
Rear panel, still with empty holes for preamp connections:
Except for removal of the tall Mosfet heatsinks, relocation of the transistors to the chassis sidewalls and reuse of the speaker posts I tried to leave the Zenductor 2s unmodified, and make the assembly as modular as possible. DC power and RCA input connections are via connectorized jumper cables. A preamp mounted to the front panel can be optionally looped into the signal path via rear panel RCA input and output connectors. I integrated two digital bias current displays and a large illuminated push button power switch into the front panel.
Heatsinks and transistors after removal:
They look a bit sad, with their legs cut off. Will they ever run again? ;-)
The amps just fit with the jumper cable connections:
Transistors mounted on sidewalls (1.4 K/W heatsinks, tested in https://www.diyaudio.com/community/...galaxy-1gx388-2u-chassis.426972/#post-7997356):
A quick test to see whether the Mosfets survived:
The 3-digit display shows the bias current. All four transistors survived, and get warm, but not hot. Definitely an improvement over the original heatsinks.
A thermal image:
Lots of filing for the bias display cutouts:
It's been a while since I had to do 'precision filing'; maybe I'll fill the gap later.
Almost finished (except for the preamp connections):
With top cover:
Rear panel, still with empty holes for preamp connections:
Attachments
Last edited:
The preamplifier (from ebay, using two NE5532 op-amp chips) is all wired up and works. As hoped, it puts out a little bit more than my Bluetooth receiver. For some reason the volume when streaming from a Macbook was insufficient. Must be an Apple thing?
The preamp gets looped in via two external 8 inch RCA jumper cables, which makes it easy to remove it from the signal chain.
Unfortunately one of the green 3-digit displays already fried itself and had to be replaced. At some point I'll probably switch the other display to yellow as well.
A recalibration of the voltmeters is probably in order as well.
The preamp gets looped in via two external 8 inch RCA jumper cables, which makes it easy to remove it from the signal chain.
Unfortunately one of the green 3-digit displays already fried itself and had to be replaced. At some point I'll probably switch the other display to yellow as well.
A recalibration of the voltmeters is probably in order as well.
IR images after bias adjustment to 1.13V over the inductors. The current sharing on the left side is perfect, on the right side the front transistor runs hotter.
Overview:
Left outside:
Right outside:
Right Mosfets:
Left Mosfets:
Front view, with both bias displays now in yellow:
Final wiring:
I moved the bias display ground reference to the inductor, to get rid of a 30mV offset. The cheap displays came pretty well calibrated (within 2%), contrary to what some of the reviews (on Amazon) said. Gluing the top panel nuts in place in the sidewall channels makes replacing the lid easier.
Overview:
Left outside:
Right outside:
Right Mosfets:
Left Mosfets:
Front view, with both bias displays now in yellow:
Final wiring:
I moved the bias display ground reference to the inductor, to get rid of a 30mV offset. The cheap displays came pretty well calibrated (within 2%), contrary to what some of the reviews (on Amazon) said. Gluing the top panel nuts in place in the sidewall channels makes replacing the lid easier.
Last edited:
Parts used:
To integrate the preamp:
- Burning Amp 2024 Zenductor 2 (pair), came with 16V switching power supplies (hopefully available soon in the diyaudiostore?!)
- Modushop Galaxy 1GX388 2U 330mm by 280mm by 80mm chassis https://diyaudiostore.com/products/galaxy?variant=12174834308
- 30mm double pole latching push button switch with ring light https://www.amazon.com/dp/B0CNYZ56ZG
- 5.5/2.5mm (NOT 2.1mm) DC power sockets & pigtail cables (2 of 3 sets used) https://www.amazon.com/dp/B0B6FFN4V5
- RCA feedthroughs (2-pack) https://www.amazon.com/dp/B004M4B7WA
- digital 3 wire 3 digit volt meters (5 pairs, 1 pair used) https://www.amazon.com/dp/B0BG31FBX5
- rubber feet (4-pack) https://www.amazon.com/dp/B08FD8TRPD
- RCA jumper cables, 6 inch, right angle to straight (pair) https://www.amazon.com/dp/B093PMLPPM
To integrate the preamp:
- NE5532 stereo preamp, single supply voltage https://www.ebay.com/itm/176909832601
- volume knob (pair, one used) https://www.amazon.com/dp/B0B375DC6Q
- insulated RCA sockets (8-pack, 4 used) https://www.amazon.com/dp/B08ZLQHGNS
- RCA jumper cables, 8 inch (pair) https://www.amazon.com/dp/B08PVSSPVM
Attachments
Thanks Dirk, it's the Bluetooth output. I think this Macbook (air, somewhat recent, i.e. a few years old) does not even have a headphone output. I did not dive in to see whether there are any output or driver settings associated with specific connected devices. I assume it's similar to Windows? It's my wife's macbook; it's been a while since I used a mac myself. I know my Windows tablet 'thinks' the DS220 bluetooth receiver are headphones, at least as far as the device name is concerned. But changing the category name does not seem to have an effect on that machine. I don't remember how to do that (list hardware devices and their drivers) on a Mac. For what it's worth, the DS220 does display maximum volume setting, as commanded by the mac, when the volume is still significantly lower than from my Windows tablet. I guess I also don't understand how volume is set and transmitted via Bluetooth. I would have naively assumed that the transmission uses the full dynamic range of the bit pattern, and volume reduction only takes place at the receiving end. But clearly the receiver thinks it's getting the full volume command, so the encoding must already be at a lower level, i.e. some of the dynamic range available (bit depth) is wasted in the transmission?
In any case, it's good to have a volume knob, and as far as I can tell the NE5532 is a very old but nevertheless competent audio amplifier. Of course there could be mistakes in the preamp design or layout, but nothing obvious that could hear. I guess at some point I will have to break out the old USB scope and look for myself...
In any case, it's good to have a volume knob, and as far as I can tell the NE5532 is a very old but nevertheless competent audio amplifier. Of course there could be mistakes in the preamp design or layout, but nothing obvious that could hear. I guess at some point I will have to break out the old USB scope and look for myself...
Some measurements with the Digilent Analog Discovery 2 USB scope (with BNC adapter) and the Audio Analyzer Suite (and also a test of pasting from OneNote) - ooops, unfortunately that doesn't work. Can't even post a single image. The formatted text works ok, though, but of course is not very useful on its own.
Direct pasting from a screen shot seems to work though.
This is the Zenductor 2 right channel THD+Noise as function of frequency, at 600 mV RMS input amplitude, and ~1.5V (0.19 W) output into an 8 Ohm resistor, with the bias voltage over the inductor at 1.22 V.
THD+N vs power:
Direct pasting from a screen shot seems to work though.
This is the Zenductor 2 right channel THD+Noise as function of frequency, at 600 mV RMS input amplitude, and ~1.5V (0.19 W) output into an 8 Ohm resistor, with the bias voltage over the inductor at 1.22 V.
THD+N vs power:
Last edited:
2nd try:
paste, then copy, delete and re-paste the screenshots one by one....
ooof, that worked!
Zenductor 2 w/o NE5532 preamp, left channel unless otherwise mentioned:
paste, then copy, delete and re-paste the screenshots one by one....
ooof, that worked!
- AD2 Setup:
- W1 out via BNC to RCA cable into left preamp input
- Preamp left RCA output via RCA to RCA cable and RCA cutoff cable adapter (loose ends) to scope probe in CH1 BNC adapter input
- NE5532 preamp Noise RMS measurement:
- (input terminated/loaded by output of AD2 generator and BNC to RCA cable)
- Minimum volume setting 1.73 mV
- Center volume setting 3.0 mV
- Maximum volume setting 5.5 mV
- Remeasure with input shorted ?
- NE5532 preamp gain:
- Input 100 mV amplitude sine (RMS = 70.7 mV)
- 9 o'clock volume knob: gain 0.33
- 10 o'clock volume knob: gain 1
- Center volume amplitude 304 mV, RMS 216.74 mV
- Gain 3.04 = 9.66 dB
- Max volume amplitude 694 mV, RMS 491.5 mV
- Gain 6.94 = 16.83 dB
- Preamp clips at output amplitude of 4.0 V
- Noise at center volume setting:
- For input of 1 V amplitude
- Output = 3.04 V amplitude
- Noise = 3.0 mV RMS = 3*sqrt(2)=4.24 mV amplitude
- 0.00424/3.04=0.00139 = -57 dB
- For input of 1 V amplitude
- DS220 Bluetooth receiver output
- White noise at maximum volume setting from Surface 9
- Used https://www.audiocheck.net/testtones_whitenoise.php (uniform and gaussian give same RMS)
- Also: https://mynoise.net/NoiseMachines/whiteNoiseGenerator.php?c=0&l=1821242730343842465000
- 600 mV RMS = 848 mV amplitude
- White noise at maximum volume setting from Surface 9
- DS220 Bluetooth receiver output
- AD2 loopback via scope probe:
- 600 mV RMS 1kHz test tone: -84 dB H2, -73 dB H3
- THD + N: -63.5 dB at 100 Hz peak (0.06666%), -67 dB in flat part (0.04%).
- THD+N vs frequency 0.0816% = -62 dB at 1 kHz
- 1 kHz spectrum:
- Spectrum and THD+N vs frequency agree on -72 dB for H2 (=2kHz) at 1kHz
- Conclusion from comparing AD2 loopback and NE5532 measurements:
- The AD2 is not much better than the NE5532 preamp, i.e. there is a large (almost equal) contribution from the AD2 to the NE5532 measurement!
- NE5532 preamp frequency response:
Zenductor 2 w/o NE5532 preamp, left channel unless otherwise mentioned:
- Spectrum for 1kHz 600 mV input signal at 1.20 V bias voltage:
- odd bump at 16 kHz ? Note: measured before changing bandwidth to 65 kHz.
- THD+N vs frequency at 600 mV input, output into 8 Ohm resistor:
- Bias voltage reading up to 1.6V during sweep at low frequencies, 1.22V above few hundred Hz: 1%, mostly H3, at 30 Hz!
- Gain: 0.6V RMS input, 1.5V RMS output = 0.19 W into 8 Ohm output: gain factor 2.5 = 8 dB
- 20 Hz: 1.35 V RMS = 0.17 W
- 100 Hz: 1.479 V RMS = 0.18 W
- 1 kHz: 1.514 V RMS = 0.19 W
- 10 kHz: 1.520 V RMS = 0.19 W
- Left channel THD+N vs frequency 65kHz bandwidth 600 mV input:
- Left channel THD vs power into 8 Ohm, sweep from 1mW to 10 W: 3.4% at 10W, 1% at 7 W
- Right channel THD+N vs frequency 600 mV input, 65 kHz bandwidth:
- Right channel THD vs power; max bias voltage reading 2.57V near 10W
Last edited:
Summary of previous post:
The AD2 seems to be barely better than the NE5532 preamp, with both in the mid -60 dB SINAD range. 14 bit dynamic range corresponds to 84 dB, so there must be other noise sources.
Also, in the early distortion vs frequency measurements I mistakenly left the bandwidth set to 20kHz, so above 6.7 kHz the third and then above 10 kHz all harmonics disappear, which shows as an apparent (but non-existent) decrease in distortion.
The Zenductor 2 does put out the promised 2nd harmonic, mostly in the permille range, except at the low and high end. It crosses over the 1% threshold below 30 Hz and above 7 kHz, at 0.2 W output into 8 Ohm.
And most importantly, both the preamp and the Zenductor 2s did survived the measurement 'torture'. I was worried in particular about the frequency sweeps, but also about the power sweep with sustained multi-Watt output over several seconds.
The AD2 seems to be barely better than the NE5532 preamp, with both in the mid -60 dB SINAD range. 14 bit dynamic range corresponds to 84 dB, so there must be other noise sources.
Also, in the early distortion vs frequency measurements I mistakenly left the bandwidth set to 20kHz, so above 6.7 kHz the third and then above 10 kHz all harmonics disappear, which shows as an apparent (but non-existent) decrease in distortion.
The Zenductor 2 does put out the promised 2nd harmonic, mostly in the permille range, except at the low and high end. It crosses over the 1% threshold below 30 Hz and above 7 kHz, at 0.2 W output into 8 Ohm.
And most importantly, both the preamp and the Zenductor 2s did survived the measurement 'torture'. I was worried in particular about the frequency sweeps, but also about the power sweep with sustained multi-Watt output over several seconds.
Last edited:
Just a quick cautionary note, in case anybody is looking at these measurements and graphs: the noise bump around 16kHz that shows up in some measurements above is definitely an artifact of the AD2 or the Audio Analyzer Suite. The wide bump moves to a higher frequency (~30kHz) when changing the upper end of the frequency scan range from 20kHz to 50 kHz.
I also saw some odd behavior with the noise floor jumping +50dB, and then jumping back down a few repeat measurements later.
Also odd: from the spectrum analyzer measurements with no test signal injection, the noise floor of the DS220 Bluetooth receiver and of the NE5532-based preamp is claimed to be the same as the AD2 noise floor itself, at -110 dBu (i.e. with 0 dB = 0.775 V RMS). I'm not sure I believe any of those numbers. I need to remeasure with the Digilent WaveForms software, and maybe also independently with REW with the Scarlett Solo external sound interface.
I also saw some odd behavior with the noise floor jumping +50dB, and then jumping back down a few repeat measurements later.
Also odd: from the spectrum analyzer measurements with no test signal injection, the noise floor of the DS220 Bluetooth receiver and of the NE5532-based preamp is claimed to be the same as the AD2 noise floor itself, at -110 dBu (i.e. with 0 dB = 0.775 V RMS). I'm not sure I believe any of those numbers. I need to remeasure with the Digilent WaveForms software, and maybe also independently with REW with the Scarlett Solo external sound interface.