Time to DIY like it's 2009! (hobby has changed, man)

Discussion in 'DIY' started by Beefy, May 29, 2021.

  1. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    Neat!

    What sort of currents are you running? I've always been curious just how much you can actually achieve through paralleled devices before they melt.
     
  2. peef

    peef Friend

    Pyrate
    Joined:
    Oct 5, 2015
    Likes Received:
    311
    Trophy Points:
    63
    Location:
    Montreal
    I'm running them right around 1.5A (per channel, of course) using the PNP circuit. This basically turns the internal pass transistor into a Darlington-- which is why the dropout voltage doubles. With the extrnal pass transitor (either NPN or PNP), I think you are limited only by transistor's SOA and by the regulator's ability to drive the transistor base.

    I haven't tried the parallel regulator method, but there is some current limiting baked in, so four of them won't get you past 0.8A. If you want to go down that route, I'd look into the LT3045, which is basically the same part, but rated for 500mA each. It's a bit more cost effective.
     
  3. m17xr2b

    m17xr2b Friend

    Pyrate Banned
    Joined:
    Mar 26, 2016
    Likes Received:
    3,988
    Trophy Points:
    93
    Location:
    United kingdomland of fish and chips
    At one point bypass capacitor thoughts start to creep in. I find bypassing situational, rarely does nothing on a high end amp, mostly worse sometimes better. Even the same caps combo sound different in different applications.

    Adding some fancy spices to a hamburger:
    [​IMG]

    I bypassed all the signal caps in the OTL amp, it does make it wagyu but it's still a hamburger.

    I got pretty much as expected, some refinement, interesting no treble timing issues so not fuzzy. Wood instruments sound very woody, surprised the bass is punchier, even smoother than before, perhaps the best relaxed listening so far, more so than an LG

    And this is my beef with fancy cap bypasses in tube amps, there's usually better value, change that IDHT to a DHT, review grounding, biasing scheme, psu damping and filtering etc.

    Ending the OTL experiment on an obvious conclusion, you can't make the crack into a teton with components. It never has that lightning fast response or immediacy the Teton offers and that's because of the operating point.

    A 6080 at super low current won't have the same gm or plate resistance as one at 100mA. Thus even if both the Crack and the Teton quote an output impedance of 150ohm I suspect the Crack is higher than that and makes for the slight mushy sound in the low end. Typically tubes are ran between 60% to 80% of their dissipation, the crack is 20% or so, great as an mid level amp and they'll last forever but it'll always be gimped for high end considerations.
     
    • Epic Epic x 5
    • Like Like x 1
    • List
    Last edited: Oct 26, 2021
  4. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    Yeah, you've certainly made that meatball as spicy as it can get!

    You ever used the 5998? I found it quite a bit snappier than the 6080 and 6AS7 variants. Z-out should be 30-50% lower, Bottlehead claims down to 80R.

    Also, I'm trying to remember the currents running through these tubes. I recall with the resistor load it was about 35 mA per section, with the Speedball the CCS was set to about 30 mA.....?
     
  5. m17xr2b

    m17xr2b Friend

    Pyrate Banned
    Joined:
    Mar 26, 2016
    Likes Received:
    3,988
    Trophy Points:
    93
    Location:
    United kingdomland of fish and chips
    I have a bunch of 5998, bass is definitely tighter and better behaved, it just lacks the mids I need to have in my life.

    It's about 33mA with a resistor for bias and 30mA with the CCS. Still, I think tubes sounds better ran closer to the intended design, like all those 300Bs with 2.5A heaters and advertised as 2A3 flavours, rarely I've seen anyone like them in a 2A3 amp. I even tried a 300B at the Stratus operating point, it sounded half asleep.

    I'll try at 75mA, see if it likes some fire under its ass.
     
  6. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    I must say though, while it might sound a little better when driven harder, I do love the tube life when operated conservatively like this. I picked up a matched pair of 5998 for something like US$40 right when the Crack was released, before the massive explosion in popularity and price of this tube. I easily have in excess of ten thousand hours on the first one, and it's still going strong.
     
  7. m17xr2b

    m17xr2b Friend

    Pyrate Banned
    Joined:
    Mar 26, 2016
    Likes Received:
    3,988
    Trophy Points:
    93
    Location:
    United kingdomland of fish and chips
    Yes, 100% on tube life and why I'm hesitant to run the tubes hard. 6080 normally have brutal short lives in PSUs, up to 125mA. Power consumption is another issue, a crack goes up to 30W, even with a tube rectified PSU and the BL63 input it's just 40 glorious watts.

    That's why when I'll change it for more current I'll add a switch to increase the cathode resistor and have a high-low current setting. A turbo button is what I'm looking for really and switch between 35mA and 75mA, even on the fly it may work.

    Obviously I'll have to give up the DC coupling but a plan is already in place, what's better than a piece of wire? No wire at all.
     
  8. peef

    peef Friend

    Pyrate
    Joined:
    Oct 5, 2015
    Likes Received:
    311
    Trophy Points:
    63
    Location:
    Montreal
    Prototyping a new Stax amp...

    [​IMG]

    It's the latest take on the "solid state triode" amps I've been playing with, and my first time trying transformer coupling to Stax. Measures well, should be making music in a couple of weeks.
     
    • Like Like x 4
    • Epic Epic x 2
    • List
  9. peef

    peef Friend

    Pyrate
    Joined:
    Oct 5, 2015
    Likes Received:
    311
    Trophy Points:
    63
    Location:
    Montreal
    It is making music. I am pleasantly surprised that the output transformer thing works at all.

    [​IMG]

    And doing a very good job of it at that. I had the opportunity to listen to a friend's CRBN and HD-60 on it, and I don't think either of us have heard sound this good out of these drivers. It's interesting that while the SR-X that I'm comparing it to has significantly higher voltage rails and more current in the output devices (EL34s), this guy sounds like the cleaner and more powerful of the two, particularly in the lower registers.

    I'm rolling the input transformers now-- I tried the Cinemag 15/15b, and since taken them out. :) The amp now runs some vintage Hammonds with are more to my taste. I'll be popping in Lundahl 1544As once I get boards made for those.

    The bass was a bit boomy when running it off a CLC supply, but adding the shunt regulators really improved detail retrieval and bass impact without it being overbearing. Even the small power supply bypass caps (the big WIMAs) made an audible difference. I'm always amazed at how big an audible difference the regulators make.
     
    • Like Like x 3
    • Epic Epic x 3
    • List
  10. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    What's the difference with an output transformer and those energizers used to connect estats to speaker amps?
     
  11. peef

    peef Friend

    Pyrate
    Joined:
    Oct 5, 2015
    Likes Received:
    311
    Trophy Points:
    63
    Location:
    Montreal
    Ha, good question! I'm using them as "plate" loads, which isn't really an option with the energizers. This allows the amp to swing well above the positive rail voltage. They are also 1:1ish (in fact 2+2:2.25+2.25), which means that the amp sees a load that's more or less equal to the driver capacitance + any transformer parasitics. The energizers are in the 1:25 or 1:50 range, such that the driver's capacitance can become significant for a low voltage, high current amp, especially when global negative feedback is involved.
     
  12. m17xr2b

    m17xr2b Friend

    Pyrate Banned
    Joined:
    Mar 26, 2016
    Likes Received:
    3,988
    Trophy Points:
    93
    Location:
    United kingdomland of fish and chips
    Sometimes DIY seems to take the expected time for the first 90% and two times expected time for the last 10% after a delay of a few months.

    The PSU of the ultimate daily headphone amp is finally done and tested. One enclosure, 48 pounds, dual tube high voltage rectification and four low voltage supplies, all with inductor full wave inputs.
    [​IMG]

    Everything overspecked to have it running cold for continuous operation. Kemet ALS industrial PSU caps for low tension supply.

    [​IMG]

    High voltage is with NOS dual C core oil sealed iron, kelvin film cap.

    [​IMG]

    Running and tested, 344V without a load, close to simulation numbers. I'll add some nice panels once the build is fully done.
    Rated at 250W and the amp will draw 75W total.
     
    • Epic Epic x 6
    • Like Like x 4
    • List
  13. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    Holy shit. So much awesome there, but what stands out most is your immaculate wiring. Such awesome attention to detail!
     
  14. m17xr2b

    m17xr2b Friend

    Pyrate Banned
    Joined:
    Mar 26, 2016
    Likes Received:
    3,988
    Trophy Points:
    93
    Location:
    United kingdomland of fish and chips
    Signal amp progress after two days of work:
    [​IMG]

    A reinvigoration of my original design, 01A with gyrator load into PX4 parafeed, this time with the knowledge and experience of half a dozen builds.

    [​IMG]

    This will run cold and far away from the PSU, complete dual mono construction and as compact as can be without clutter or compromised.

    [​IMG]

    Going away from the chassis norm, the tubes are attached from the bottom so it can be powered without a top plate for easy tweaking and measurements. Full Magnequest unobtanium iron, nickel output, oversized chokes, all high end tried and tested PSU caps. Built with the main focus on PSU, the heart of any good amp.

    [​IMG]

    After four days of work I still have to do some casework before soldering can begin, one side if off by 3mm and have to undo, drill and put back together again. Still, it's nice to see how it will look completed, good motivation to keep going.

    The amplifier gods requested a sacrifice, a good sign as all my best amps were built in escaping vacuum.
    [​IMG]
     
    • Like Like x 5
    • Epic Epic x 3
    • List
  15. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    I finally had some free time over the holiday break to finish up a major DIY project, a re-kejiggering of my big DIY DAC!

    First some history...... ESS released their first Sabre DAC chip in the heady days of DIY in 2008. Even though history hasn't been kind to the ESS9008, it was something completely different at the time and really generated a lot of excitement. The very first availability of this DAC came in DIY form, the Buffalo DAC from Twisted Pear Audio. I was a relative DIY noob at this point but jumped in really early (serial # 4!) and managed to create a basic but functional DAC. It had switchable COAX and TOSLINK input, TPA's LM317/337 power supplies, and an IVY linestage.

    [​IMG]

    As 2009 rolled round I had much improved DIY skills, and completely rebuilt this DAC. With a nicer case and FPE panels it certainly looked the part, but the best bits were all inside. I added AMB's E24 power switching circuit, TPA's CS8416 based SPDIF MUX module, AMB γ1 USB input, TPA Placid BP power supply for the linestage, and custom magnetically-shielded transformers from SumR (30VA 9+9V for digital, 30VA 15+15V for linestage). I later replaced the IVY with an IVY III. It was all just about as good as you could do with DIY at the time, unless you literally built everything yourself from scratch. This served as my main headphone DAC for about five years, and when I stopped using headphones it got constant use for 7 years as the DAC for my TV and speakers. A true workhorse.

    [​IMG]

    [​IMG]

    [​IMG]

    So now, revisiting this in 2021...... I thought this would be a no-brainer upgrade. Just buy whatever the newest module is from TPA, new USB input board, pop them in, and I'm back on the cutting edge. Imagine my surprise when the ESS DACs have really fallen out of favour, and that my prized input scheme is completely outdated. These days, it is all so easily replaceable with a Pi, Pi hat, and even an OLED touchscreen.

    Deciding how to approach this upgrade took a lot of thought (read: angst). Originally I thought of dropping in an R-2R module to be one of the cool audiophile kids, but decided it was best to make that an entire new DAC (which I did!). I thought about different upgrades to modernise the functionality, but this would have meant expensive new front and rear panels. I didn't want to completely give up multiple inputs, yet keeping the outdated CS8416 module in the chain by default seemed like a performance bottleneck - a USB module straight to I2S is way better than USB to SPDIF to CS8416 to I2S. What to do?

    In the end, I took the easy and boring way out and basically followed my original plan. But I also took an each-way bet with the input switching, and here's how it turned out.....

    [​IMG]

    The DAC itself is the TOTL ESS9038 Pro module, with the Mercury linestage and Trident SR regulators. Two Centaur power supplies feed the digital side, with a Placid HD BP for the linestage. USB input is a JLSounds I2SoverUSB v.III as used in my other build, connected to the DAC by I2S. In essence, this is currently set up as a dedicated USB DAC.

    But the input switching is all there, powered up, and SPDIF lines connected (including SPDIF from the USB module). All it needs to turn back into a multi-input DAC is to move the I2S lines from the USB module to the CS8416 module. I think it is an elegant way to get the absolute maximum performance now over USB, but still allow other inputs later without any new modules or casework. All the extra wiring from the inputs and digital power makes it a bit of a rat's nest, but with the lid on you'd never know.

    All things considered, I'm really very happy with how it turned out. The output voltage as configured is very hot at ~3 VRMS, so I've set a fixed digital attenuation to ~1 VRMS. Early listening tests are very positive; noise floor is dead silent into my M3/Clear combo, everything is very dynamic yet musical, and I notice none of the etch/glare/brightness complaints other have reported with ESS DACs. My only concern with the whole build is some transformer mechanical hum because my mains voltage is 122V. But this will ultimately be relocated to my office work where I get a solid 114 V, which should be better. At the office will feed my Exstata/Lambda and Crack/HD650.

    Looking at this now, this DAC still has an original part from 2008 - the IEC inlet! Unfortunately I have to replace that as well, because the plastic housing is cracking. But wow, did I ever get awesome value from that original ESS9008 module.....
     
    • Like Like x 6
    • Epic Epic x 3
    • List
  16. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    Starting to wrap up this little journey with a few final tweaks on my builds...

    First up, some changes to the M3. My 'low' mains voltage of 115 VAC at work (vs 120-125V at home) highlighted that using a 12+12 secondary transformer to generate 27 VDC is really pushing the limits of the S11 power supply. So I ordered a new custom unit from SumR; 30VA, 115+115 primary, 13+13 secondary, electrostatic screen between primary and secondary windings, fully magnetically shielded. The 2 VAC voltage bump seems small, but gives me an extra ~1.5 VDC pre-regulator and has allowed me to bump the output by ~1.5 VDC. A small change, but worth it for both better regulation and more voltage swing from the amp.

    When ordering that transformer, I also ordered a custom unit for the R-2R DAC. The two smaller transformers I originally installed over-volted stupidly, creating way more heat than was necessary. So I worked with Richard to replace these with a single unit that would better hit the ideal voltages. It ended up being 40 VA, 115+115 primary, 11+11+7+7 secondary, electrostatic screen between primary and secondary windings, fully magnetically shielded. The design hit the target voltages perfectly, dropping the total heat output of the regulators from ~2.8 W to ~1.7 W, while still having plenty of regulation headroom. I now feel a lot more comfortable about the longevity of the power supplies.

    [​IMG]

    Last big change was to the volume control in the Exstata. It originally had a dual PGA2320 volume control that served me very well for many years. But the PGA2320 design is almost 20 years old - it uses an internal resistor network to set the gain of an OPAMP, and it needs a really good analog power supply. So I dug up some boards for Kevin Gilmore's 8-bit latching-relay shunt attenuator that I bought as part of a group buy back in 2011 and never assembled. I built it with an input impedance of 25k per phase, 0 to -127.5 dB attenuation in 0.5 dB steps, using Panasonic ERA 0.1% 25 ppm SMD resistors. Two boards for balanced. It's controlled by an Arduino UNO with 10k linear pot, and powered by some spare LM317-based power supplies and transformers I had in my parts box. I never planned for such a big complicated multipart setup so layout was a challenge, but I managed to fit everything in without compromising my dirty AC and clean signal 'zones'.

    [​IMG]

    [​IMG]

    It works really great, and channel matching measures literally perfectly. But there are caveats. I suck at coding, so I am entirely reliant on other people's code, and the code I am using fully turns up the volume control on power-up. That's not great. But so long as I don't power up with phones connected, it should be fine. Second, the attenuation is actually too good, because the range from -127.5 to about -70 dB provides so little signal. Only the top half of the volume control is usable, which is quite odd versus normally having only the bottom half being usable! But the code I'm using does allow for remapping the attenuation curve, so I'm just going to have to suck it up and learn to code it.

    Last little issue is that the Exstata has been moved to work, where the line voltage is lower than I designed the transformer for. The SSR I use for power switching eats an additional 1.5 VAC from this. All things said, I only have about 13 VDC dropout headroom for 300 VDC in the power supply, which is simply too low. But there's no way I'm ordering yet another custom transformer! So, the SSR is being switched out for a standard mechanical relay that has no voltage drop; it seems small, but +1.5 VAC on the primary becomes +3.1 VAC on the secondary, which becomes +4.5 VDC. That takes pre-regulator headroom to 17.5 VDC - maybe still not ideal, but a big money saver. With that change, it should complete the work rig alongside the Buffalo and the Crack...

    [​IMG]
     
    • Like Like x 4
    • Epic Epic x 3
    • List
    Last edited: Feb 17, 2022
  17. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    7,467
    Trophy Points:
    113
    Location:
    Winnipeg
    Ah yes, it feels so good to finally use something from the DIY drawer of shame...
     
  18. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    I felt very little shame for those boards, they were dirt cheap at the time.

    I almost feel more shame having built the thing! Spending a couple of hundred bucks on parts, to replace a volume control that is already better than the majority of stuff on the market, with something that is probably not audibly better, and actually functions worse in some respects? Yeah, that sort of nervosa should bring shame.

    But, the clickety clack clack clack of 16 relays is pretty damn funny......
     
  19. Cspirou

    Cspirou They call me Sparky

    Pyrate
    Joined:
    Sep 27, 2015
    Likes Received:
    8,200
    Trophy Points:
    113
    Location:
    Northwest France
    Any particular reason you didn't go with the 4 channel PGA4311?
     
  20. Beefy

    Beefy Friend

    Pyrate
    Joined:
    Apr 10, 2021
    Likes Received:
    1,738
    Trophy Points:
    93
    Location:
    Canada
    In a word, convenience. The board designer (luvdunhill of Head-Fi and Head-Case) had created a really nice design with PGA2320. And I went with his volume control board over other options, because I was also buying machined Teflon Stax jacks from him. I think he even sent me the 1000V hookup wire I used in my Exstata in the same package.
     

Share This Page