DIY talk

That's my initial reaction, too. But! easily solved by smooshing some rope caulk into the steps and smoothing it out with wet fingers.

(Yikes. smooshing and smoothing caulk with wet fingers. I'm sure I set off some kind of alarm there...)

 

Actually just remembered this mini-tutorial from the TotalDAC guy

http://vincent.brient.free.fr/round_horns.htm

No lathe but requires a lot of rasping and sanding. I’d be a afraid of making it uneven

 

Just get some Play-Dough and fill in the ridges (if you're looking for easy-peasy). Otherwise, plenty of options - expanding foam, paper mache, resin, concrete, etc.

 

For a surface to diffract incoming/passing saves, it must be on the same order of magnitude as the wavelength. In the case of a backloaded design, your steps are in fractions of an inch and the wavelengths are in feet, so there's nothing to worry about.

With a front horn though, you're dealing with frequencies who's wavelengths are in the same range as your steps, so that's a problem.

For easy-peasy, I'd recommend plumber's putty mostly because it does not shrink/crack/etc as much as the others:
https://www.homedepot.ca/product/oatey-9-oz-stain-free-plumber-feet-s-putty/1000723576

 

I just saw that Laserboost (euro based) is offering sheet metal bending now.

This is great and would allow me to design L-bend (schiit style) or U-bend sort of chassis. Would simplify grounding a bit when it’s all the same piece of metal

 

https://www.systemthree.com/products/sculpwood-max-putty

 

I made a few small updates to the sch for the current/voltage amp.



The board will now support parallel input JFETs. A lot of the FETs that I like will bias to just -0.3V at currents that we need. By putting two in parallel, the current in each device is halved, and the buffer's overall transconductance goes up. This lets us use parts like the 2SK3557 or NSVJ3910 (which might be better than the 2SK170, even).

Secondly, the servo got much more complicated. I didn't like how the previous servo seemed to impact the overall gain with certain resistor values, which in turn reduced the servo's authority. This new version uses the servo op amp to generate a reference current for a complementary current mirror, which shifts the bias of both gain transistors to zero out any offset.

The current mirror uses some very cool, inexpensive matched BJTs: the NSS40301 and NSS40300. They are no SSM2212, and their capacitances are a bit higher than I'd prefer, but they are a perfect fit for this application. I also think it's funny that they look like op amps.

 

The amp boards are done. This was a challenging layout!




Some features worth pointing out--

The board is designed to mount directly to the heatsinks of a 2U chassis, with connectors for the power supply, input, and output. The conectors are Molex Mini Fit Juniors, which were selected mostly because the name is silly.

There are lots of options for bypassing the power supply rails, which physically span the top and bottom of the board: film box cap, MLCCs, and the SP-Caps that may or may not be killing GPUs. The best option or combination will ultimately depend on the type of PSU or regulator being used. The SP-caps in higher voltages (>5V) are getting NRND'd, but offer such extraordinarily low ESR/ESL that it's worth considering for builders who have access to them. I've been using them for a while and they have not crashed any of my amps.

Lots of resistors were replace were replaced with series strings. In the case of the feedback resistors, this was done to reduce resistor distortion, as these see the full output voltage swing. Resistor distortion is likely not robbing anyone of good sound, but the series resistors is more or less a free lunch. In the case of the servo, this reduces leakage, which may or may not be a big deal given the very high impedance of this section of the circuit.

Lastly, for folks with a servo allergy who are interested in building the voltage amp version, there is a jumper to disengage the servo. The transconductance amp will not bias correctly without a servo. With dilligent trimming, offset should be below a few mV.

I'd like to build this myself before releasing the gerbers to the community, but if anyone would like to build along, they're welcome to shoot me a pm.

 

Always game

 

What are you thinking this will need in terms of power supply requirements? Is something like an AMB S22 going to be sufficient? Ultra-low noise linear regulator? Shunt?

 

Depends if you want to power speakers or cans. The general reco would be something that can put out +/- 12V to 15V, and depending on whether or not you're planning on driving speakers, should be able to supply at least 1A peak, 200-300mA quiescent current per channel. The op amp servo runs off of the main amps supply, so going to a higher voltage PS will require a FET-input op amp that can handle the voltage, such as an OPA445. That said, the amp cannot swing rail to rail, so on a good day 12V rails will probably get up to 18Vpp on the output due to the cascoding of the output stage.

AMB S22, KG's Goldreference, or a super reg would all be good choices. I'll be doing a bipolar version of the LT3042 booster using the LT3045 and LT3095, but it's probably a good way off. Normally I'd recommend a shunt, but you will need to waste a lot of idle current as the amp can exit class A.

The simmed PSRR is around 50dB-- good but not extraordinary-- so it's worth using a good sounding supply. If you hate good sound, an LM317, or an unregulated supply will technically work.

E: incidentally, does anyone know of a good source for inexpensive aluminum boxes that can hold a 2U heatsink? Something like a mini Dissipante looks good, but isn't free.

 

Yeah, I was musing at how small I might be able to build one of these in a Mini Dissipante for headphones. The dual rail LT3045 module I used in my DAC is a tiny little thing that should easily be able to handle continuous 200 mA per channel at 15 V, if the transformer voltages are carefully controlled.

Might be able to squeeze it all in a 230 mm wide, 250 mm deep Mini chassis. But yeah, the price is OUCH, definitely interested in potentially less expensive chassis options!

 

Modushop sale until July 18th.
ESTATE10 for 10% off for 50 euro min order
ESTATE15 for 15% off for 150 euro min order
ESTATE20 for 20% off for 250 euro min order

Maybe save a few bucks

I looked around before, didn't find anything cheaper then Modushop for a 4U 400 Dissapante I bought late last year.

See some stuff on ebay, but shipping was killing those.

 

So I'm looking to do a boombox project with a class D module, probably TPA3116. I wanted to make it battery powered and reading the specs it requires a single supply with a range of 4.5V - 26V. It also seems to have all sorts of built-in protection. To me this seems like I can hook the batteries up directly with a switch and not have to worry about power supply failure causing damage.

Is there anything else I should consider?

 

No, I did something similar with a small 8v lead acid battery. You could add a few caps for a bit of reserve but most boards have small caps on them

 

So back when Soekris originally released his R2R DAC module Hifiduino was pretty much the defacto authority on it. Then for some reason he stopped posting for years.

Recently noticed though that he started posting again for anyone that wants to revisit

 

I’ve been looking at tube preamp schematics and they all seem pretty basic. Pot -> buffer -> capacitor coupled output.

what would an endgame preamp look like? Is transformer coupling a thing for a line stage?

 

No idea what endgame looks like, but an output transformer seems unnecessary in most cases. The buffer is usually designed to be more than sufficient to drive the input impedance of the next stage, and adding the transformer is an expense and possible degradation, not to mention it possibly takes away some or all of the gain that you wanted the preamp for in the first place.

Transformer only seems necessary if you need SE/XLR conversion inside the preamp.

 

You can do transformer coupled and TVC at the same time, so your OPT is also your TVC. Thomas Mayer does one with a 10Y. Eg here. His "endgame" preamp is differential and very expensive (due to huge numbers of silver chokes and transformers).

 

I just noticed that in the era of premium tube amps from McIntosh, Quad, Marantz, etc; they all made Class AB PP with feedback to provide as much clean power as possible. The preamps units however were simple cathode followers with capacitor coupled outputs. This seems like it would be a fairly high distortion. However since current application remains low, the tubes are never pushed into high distortion? I would think some obsessive objective-minded tube guy would have done a similar push-pull type of preamp at some point. Which would only be possible with a transformer output to convert to RCA

Totally random but I was searching McIntosh schematics and came across this article

That Time the Navy Used the Grateful Dead's Amps to Listen to Soviet Subs - Vice.com

Edit:

I forgot about the Freya. No schematics available but description says it’s similar to a circlotron. Really nothing else like it out there. Not sure why anyone hasn’t tried similar before

Measurements in Stereophile look really good with F2 at -80dB. I saw measurements on ASR too that looked like total garbage and I wondered why. Then I noticed he used an output voltage of 4V into 40Ω. For a preamp. Lol