DIY Audio General - stuff you're making, tips, advice sought, etc

I’ve got a GEC KT88 which has clearly run away and while the glass is intact there’s a hole melted in the metal anode ! Given that that will be something pretty refractory (nickel ? molybdenum ?) that takes some doing.

It’s called runaway, incidentally, because when the metalwork (the anode, mostly) gets too hot it tends to release gas, which is normally adsorbed stably on its surface. The gas is ionised by the regular electron current and the resulting (positive) ions are attracted towards the (negative) grid. Grid current therefore flows. This passes through whatever component is holding the grid down to its nominal bias voltage (often 0V, i.e. ground) and the current generates a positive voltage across that component, thereby raising the grid voltage. That switches the valve on harder. Which heats the metalwork further. Which releases more gas. Which generates more grid current. And round and round until some hard limit is hit, usually way, way beyond what the valve is rated for. And pretty soon there are tears :cry:

3 Likes

please elaborate

2 Likes

Sorry, that’s pretty much all I know :grin: (learned the hard way, given the price of old-stock M-OV KT66s :cry:).

Don’t worry though. That nice Prof Spangenberg’s comprehensive book (1948, 878 pages) is now available free online and I’m sure you’ll be able to find what you need in there (somewhere).

It’ll be something to do now the dark winter nights are drawing in.

1 Like

linky no worky, maybe this one does?

https://www.worldradiohistory.com/BOOKSHELF-ARH/Technology/Technology-Vacuum-Tube/Vacuum-Tubes-Karl-R.-Spangenberg-1948-(877-pages).pdf#search="spangenberg"

1 Like

Thanks Simon, somehow I must have cropped the link in the cut-and-paste. I (hope I)'ve fixed it now.

Not sure how you work that out.
Even if they only made 80 tubes a week it would only take 6 months to make 2000 tubes.

Takatsuki make 600 pairs of the 300B per year so that 2000 equates to nearly 3.5years.

Except they are individually numbered, so half that.

still not less than 6 months though :stuck_out_tongue_winking_eye:

Jim had his for a year!
So all works about right I reckon :grinning:

Been fannying around with my Audio Note Kits DAC 4.1 on-and-off for what’s likely the best part of a year.

Before pic:

Diverting from my usual approach (i.e. impatiently throw All The Things at it in a couple of days and then be bemused that the result isn’t all I hoped for), I’ve done the work in stages, living with and listening to the changes over a period of time to get an accurate (if subjective) idea of the effects, reversing anything that proves substandard.

As it’s an electrically-noisy brute I experimented with a couple of different onboard cartridge-type mains filters: those were both actively harmful to SQ and were removed. Instead I made an offboard filter with DC blocking and no capacitative shunts of any kind - this was a definite improvement: no more RFI/EMI beng coupled back onto the supply.
I also tried an Earth lift - again, actively harmful. Instead a simple high-current, low-L inductor now connects system Ground to mains Earth - this made a small but tangible improvement.
Both of these probably have more effect on other components connected to the same power supply by preventing noise proliferation.

The analogue output stage was pulled and the following done:

  1. board mounted on vibration-decoupling rubber standoffs.
  2. 0.1uF ANUK copper foil caps upgraded to 0.22uF ANUK.
  3. R1 and R11 metal film resistors upgraded to ANUK Tants.
  4. cheap ceramic B9A valve bases upgraded to Belton VTB9-PT.
  5. Soldered wiring connections to-and-from the PCB replaced with screw-terminals for ease of maintenance.
  6. All solder joints reflashed and board thoroughly cleaned.

This cleaned-up the overall SQ, the noise floor was reduced, and the bigger caps seemed to help with the DAC’s bass-lightness - though bigger-still 0.68uF Audyn copper caps were an overall downgrade in SQ.

The biggest lift in overall SQ resulted from changes to the PSU - something that’s always worth doing IME, but I’ve never had a result as transformative as this before:

  1. Majority of soldered wire connections to PCB replaced with screw terminals for ease of maintenance.
  2. 2 x ST L78S12CV voltage regs (higher-current 7812) for filament supply upgraded to NewClassD UWB2 Regulator Mk2 12v
  3. C9 - 2200uF 35VDC Panasonic FC changed to same at 3900uF 35VDC (not an improvement, see 10.)
  4. PSU grounding wire link upgraded to Eupen GNLM 1.5mm
  5. PSU to output stage + and - upgraded to Eupen GNLM 1.5mm
  6. OPTs grounded to new ground/Earth post via significantly shortened wires (+ wound on to a ferrite) for reduced interference pickup
  7. All wire attachments to PSU board remade with screw-clamp terminal blocks except + and - to rectifier which were resoldered
  8. All solder joints on PSU board reflashed, aside from low voltage diodes
  9. Brass PCB standoffs replaced with vibration-decoupling rubber ones.
  10. 470uF 16VDC voltage reg output cap upgraded first with Panasonic FC 100uF 63VDC - replaced in turn with Elna gold 1000uF 35VDC
  11. C9 replaced with 2200uF 63Vdc Mundorf MLytic AG
  12. C3, C5 & C6 replaced with 100uF 450Vdc Mundorf MLytic HV
  13. C7 upgraded to 220uF 500Vdc Audio Note Standard Electrolytic Capacitor
  14. V2 nasty-ass ceramic Octal valve base for 6X5 rectifier upgraded to Belton VTB8-ST-G
  15. V1 nasty-ass ceramic B9A valve base for ECL82 upgraded to Belton VTB9-PT

Changing the filament supply voltage regs caused a bit of a panic - the improvement to overall resolution was way more than I’ve previously experienced with similar mods in the past, but it also emphasised sibilant recordings to an unpleasant degree (the Magellans lurched into ‘Merciless Mode’ :open_mouth:).

Fortunately, changing the VReg’s output caps to larger Elnas, plus replacing all of the now ~14 year old smoothing caps with higher-quality equivalents returned balance to the top end.

This latter - plus changing the valve bases - lifted overall resolution way beyond what I’d expected, even more so than changing the VRegs. I was originally only changing the smoothing caps because I may as well do it now while the DAC’s in bits rather than waiting for a random failure. I also only purchased expensive foo smoothing caps as a sacrifice to my Audiophilia nervosa - I didn’t expect them to outperform the industrial stuff to the extent they do. Since all of the pulls measure 100%, it’s not like those had degraded - the sodding Mundorfs are just significantly better…

Foo works! Who knew?

The most cost-effective upgrade has to be the Belton valve bases - I wish I’d realised earlier in my tinkering ‘career’ quite how incredibly shit cheap chinese valve bases are, and how much improvement these affordable (£3-8 typically) bases make - used them in the pre to great effect, and it’s the same here.

Other miscellaneous tweaks:

  1. Cheapo BNC S/PDIF digital input socket replaced with Neutrik
  2. Badly soldered internal S/PDIF coax cable replaced with Van Damme UPLC-OFC Plasma Grade coax.
  3. All internal mains wiring replaced with Schlöder Eupen GNLM 1.5mm
  4. Furutech Rhodium IEC power inlet added
  5. Internal shielding plate added to enclosure lid above digital board
  6. USB input board removed.
  7. AES/EBU XLR balanced digital input wired-in to USB’s former input on digital interface PCB.
  8. Incorrect male AES/EBU XLR balanced digital input socket upgraded to female ‘Eizz’ PTFE / copper socket.
  9. Balanced XLR outputs wired-up and working.

Aside from reinstating the AES/EBU input mentioned, I’ve got no plans to tinker with the digital PCB - too many sensitive unobtainium parts on it, and I have very little knowledge of what would be worth changing in any case.

After pic:

Sonically this thing’s almost unrecognisable - the uplift in overall resolution is enormous, transformative, in all my 20+ years tinkering I’ve not experienced this much improvement. There isn’t much point listing the usual audiophool hyperbolic buzzwords - it’s all just better in every way.

Some of the improvement is down to better components or improved layouts, but some of it will be down to putting right some of the clumsy DIY assembly work.

Nearly forgot the most important upgrade of all:

The piss-yellow LED on the front panel is now an eye-soothing green :ok_hand:

18 Likes

2 Likes

The answer of course should be retina searing blue.

Perhaps you could have your carer read you a paragraph at a time with your nitenite Horlicks?


Jesus… I’m actually old enough to remember when blue LEDs seemed cutting-edge and a bit cule… Depressing thought OTD.

1 Like

Another critical mod - Jays Audio CDT display is now green rather than blue:
Achieved by the fierce-technical measure of removing the front panel, heating it up to the point that the egregiously strong glue holding the blue acrylic display panel screen could be jemmied free, and replacing it with a green acrylic one I made.

Moar betterer :ok_hand:

17 Likes

Madman, but then, you knew that.

2 Likes

True story :+1:

I do kinda enjoy this sort of tinkering and personalisation.

And I do fucking hate blue LEDs…

And and green sounds better, natch…

Closet Naim wannabee :laughing:

1 Like

The Nobel Prize winning technology required to create them vs the hate they receive, it’s almost Trumpian anti science level stuff… :wink:

ImOKwiththis

One of the ten best systems I’ve ever heard was all Naim electronics :ok_hand:


No such thing M8, it’s all a scam…

7 Likes