Neurochrome LM3886DR Build

Having the chip soldered into the circuit does reduce the ESD susceptibility somewhat, but not enough to rely on it to fully protect the IC. ESD requirements have been tightened by the various semiconductor manufacturers through the years. In "the old days" 2000 V human body model (HBM) was required for ESD compliance. That may sound like a lot with 2 kV applied to a low-voltage IC, but HBM is pretty easy to pass due to its series resistance and inductance. These days more stringent models, such as the charged device model (CDM) and machine model (MM) are applied. As result, more modern ICs will generally survive higher voltage zaps than older ICs. Recall the LM3886 is 25 years old.

It's easy to generate high voltages just by shifting in your chair and such. The climate in Calgary is dry, especially in the winter. Indoor relative humidity down around 15-20 % is common. So I pick up quite a bit of charge just getting up from my desk chair. For a while I would ground myself on a grounded screw by the light switch for the room, but found that rather painful at times. I decided to explore some options for how to reduce the pain, so I started adding series resistance. Once I got up around 1 MΩ, the pain level started to go down ... until the spark just jumped across the resistor instead. That's 9-10 mm of spark. Figure 1 kV/mm or about 10 kV.

My work surface has a grounded electrostatic dissipative mat on it. I maintain good contact with the mat as I'm working. I wear a wrist strap when necessary. The mat I use is 80x160 cm, so a bit pricey ($80, I think it was). You can find smaller ones for less.
All my fully assembled products are assembled professionally. The assembly guys take ESD seriously and take all the necessary precautions to prevent ESD damage. This includes ESD smocks, grounded footwear, wrist straps, and ESD safe workbenches and chairs. I would also expect them to apply ESD dissipative wax on their shop floor. We used to do that regularly in our lab at TI.

Sadly, ESD strikes can result in latent damage. The damage may reveal itself months or years later. Or it may result in reduced performance, as in your case.

Tom
Fantastic info about ESD. Big gap between what I knew and all this. 10kV from the human body makes one wonder how we're not going all sparky as we walk around, doesn't it? :D I'd have never guessed such high figures. It's also depressing because I don't really know now which of the components in my builds may have suffered latent damage.

And Gregtt, fantastic build. I've never seen so much packed into a chassis like that in DIY builds. It also gives me a lot of confidence to try the less ambitious builds I keep thinking of.
 
It's also depressing because I don't really know now which of the components in my builds may have suffered latent damage.
Yep. That's my point exactly. Note that I'm not suggesting that the average hobbyist sets up an ESD compliant lab. The cost of ESD compliant furniture, ESD wax on the floor, smocks, straps, whatnot would kill the hobby pretty quickly. But an ESD mat with a ground strap (that you connect to the building ground) is not that expensive and cheap insurance. DigiKey is a good source for those.

I would expect any production facility to have ESD measures in place and to test 'wearables' such as wrist straps and heel straps at least daily.

Tom
 
Hello LM3886DR Fans! I have enjoyed browsing through this thread as I have been building my own, picking up some nice tips here. I appreciate the knowledge exchange!

I repurposed a case (Hammond RMCV Series) used with 3 others to run a pair of Linkwitz Orion active speakers. I did this back in 2005 using Hypex UCD180HG's. I have the itch to try something different, and maybe like the greggtt build. For now just experimenting and having fun (now that I am retired) to see what makes sense. I stumbled on Tom's Neurochrome LM3886 boards - they are the best I could find, so did the LM3886DR first, and then will do the Modulus-86 next. This unit employs the DIY Soft Start and Speaker Delay/DC protect.

So here are my results, which I am very happy with. I have been playing on my current stereo setup with Klipsch Forte IV's, comparing mainly with stereo tube amps, such as McIntosh MC240 and a recent DIY build of a 2A3 SET amp. While there are enjoyable differences I believe the LM3886DR sounds great against them. I has a much smoother upper-mid and highs, something more satisfying to me than the Hypex UCD's. I am guessing Class D has improved substantially since 2005, especially reading about newer Ncore or Purifi's, but not sure I will go back.

At any rate, here are some pictures and these are my results as measured on my HP8903B. Even though the HP is much more limited (but perfect for tube amps) than the newer AP equipment, I am impressed on what I am seeing - the best I have seen on this rig. Great job Tom.

Power Rails: +/- 28VDC using Neurochrome Power-86, with Antek AS-2222 transformer.
Full Power: 38.5W (both channels very close), 0.886Vrms in.
THD+N: 0.0065% with 30K filter on
DC Offset: 4.5mV
Gain: 26dB
Signal to Noise: 97dB at full power with covers on

Will post the Modulus-86 build sometime this summer.
Cheers!
 

Attachments

  • IMG_3974.jpeg
    IMG_3974.jpeg
    636.8 KB · Views: 200
  • IMG_3975.jpeg
    IMG_3975.jpeg
    560.6 KB · Views: 195
  • IMG_4050.jpeg
    IMG_4050.jpeg
    425 KB · Views: 184
  • IMG_4047.jpeg
    IMG_4047.jpeg
    380.1 KB · Views: 169
  • IMG_3989.jpeg
    IMG_3989.jpeg
    514.6 KB · Views: 188
  • Like
Reactions: 1 user
Very cool. It looks like it came together nicely for you. Thank you for sharing. The SNR measurement is probably limited by the HP8903B.

Tom
Yes I think so. When I do a loop back on the HP I get SNR: 102dB, THD+N: 0.0018% (-95dB). I think that is fairly close to spec. This guy was designed at a different time but still working fine - I expect a re-cap in the near future - that might make some minor improvement. I have it paired with the Tek 2225 to show the HP's monitor output to reveal the signal after the fundamental is notched out - in this case not much left...
 
One other comment - I thought I was very conservative on the heatsink for 2 LM3886's but in actual use, even running fairly loud with music, they don't get that that hot - just using the feel test. That has been my previous experience but reinforced now by Tom's commentary on crest factor. For sure testing to full power with a sine wave gets this baby hot; however, in practice most of us are playing music through our system and the stress is less; hence, can get by with a smaller heatsink. Also, as can be seen by the pictures I was space limited so I guess it works out for my implementation. Size accordingly, especially if you are going to play Van Halen at 120SPL - that used to be me at a much younger age :cool:
 
For a while I would buy defective HP8903A on eBay for not that much money and repair them. Usually it was the power supply caps that had dried out. Swap those and the box came to life just fine. I'd then run them through the calibration test procedure using the calibrated instruments I had access to at the time. It's a great box for its time. I do like the instruments of that era.

My first scope was a TEK 2215. The little brother to yours it looks like. :)

Tom
 
Thanks to Tom for a great quality product that was easy to build. I took his advice and went with a dual mono set up. I know people will cringe at the absolutely useless VU meters but i couldn't resist. This is my first amp build. I built the chassis from scratch, with the heatsinks coming from a renewables company. I cut the heatsink in half to create the two sides. The amp sounds very quiet, super detailed and musical if fed decent material. Highs can be a touch forward but I put that down to the Wiim Mini and my old B&W DM14's (now being re-done to look more modern). I will get a decent dac as the one in the Wiim is not great.
Thanks for the help Tom and well done on brilliant service and product.
I will build a pair of Modulus 86's in the not too distant future or maybe save that for the autumn.
 

Attachments

  • Amp complete.jpg
    Amp complete.jpg
    376.8 KB · Views: 139
  • Chassis from front 1.jpg
    Chassis from front 1.jpg
    472.6 KB · Views: 139
  • Chassis from rear.jpg
    Chassis from rear.jpg
    420.6 KB · Views: 135
  • Like
Reactions: 1 users
Thanks, it looks better in the flesh. No issues at all with ground loops. Super quiet. The VU meters are a bit of fun, I used a completely separate power feed for them, 12v. I have the meters connected from +- on one channel output and + on the other channel. I can't here any degradation but have since took my B&W's out so I've only heard the sound on the ceiling speakers in the kitchen. I will be interested to see if the VU connection causes any difference (they are powered by a separate board)
Mike
 
The output impedance of the LM3886 is pretty low, so I wouldn't expect that you'll be able to hear any difference with/without the VU meter added.

I like the clean build as well. Nicely done. One thing you may consider is to add standoffs on the two front mounting holes on the PCBs so they aren't cantilevered off the heat sink.

Tom