Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Need help in details building Kevin Gilmore's Blue Hawaii Amp
New Posts  All Forums:Forum Nav:

Need help in details building Kevin Gilmore's Blue Hawaii Amp

post #1 of 7
Thread Starter 
Hi, I join this forum because plan to build a Kevin Gilmore's Blue Hawaii for myself. Starting out from
the original __ I have a few questions:
For replacing the 2SA1968-CCS I'm thinking about the following Possibilities (see schematic):
1a) Plain+simple: ANODE---R 10k---L (choke)---PLUS. Wehre L can be something between 0 (as prosponed in the original Document) and so that R*L goes << 20 Hz. Seriously: For me it's just simplicity at the right places which makes the difference from fancy to excellent. But: is that the right place?
1b) IXYS 10m90s (hard to get here in Germany, don't wanna wait at import customs, but maybe Mouser+Fedex have agreement so that it could work without waiting ... not sure...)
1c) Simulation AM1 or AM2 (just for orientation, T4 could be omitted) - can't find a 900V PNP (btw: 2SA1968 is documented as 
NPN what does that mean?)
1d) AM4 Simulation: Gives promising result
1e) AM4 again, but take "mu follower Output" after the 1st MOSFET to decouple capacitive load from the tube - see http://www.pimmlabs.com/web/Active_loads_and_signal_current_control.html
1f) AM5 not realistic, I think
1g) AM6 promising result



Power Supply: Is it possible to simply omit the 2SA1968, i.e. Rectifier+---Some R 30Ohm--+--Some 100W MOSFET with Gate directly 
connected to the Zehner Diodes--+--o
(+ goes to C's). I don't konw much about tubes - so:
2a) Is it REALLY somehow destructive for EL34's if anode voltage is there before they are heated up?
2b) Can one expect an audible advantage when using a 400-Volt Version of the much more advanced KGSS PS?
I mean 60 years ago they treated transistors like tubes, I don't wanna start to threat tubes like transistors now - 
maybe voltage deviations are not important...
2c) did someone REALLY find an improvement when using DC filament heating?
3) The amp will stand 10 cm aside the source, so symmetrical connection is only an option, but not crucial.
So: Replace the 2SK389 with 2 * 2SK170? They are pretty cheap, so I could buy 10 or 20 and try to pair it
DIY - but is that possible without a 6.5 digit multimeter?
Btw: I have an understanding problem: Why is symmetry so important in this case? In a "normal" B-class
Amp where each side is responsible for a half-wave that's clear. But here, as I understand, it's different: A small 
section of the diaphragm with the charge q which depends only(?) on the bias is moved by the force
F = q*E where E comes only form the voltage difference between the stators. So if the input of say 0V...500mV
ends up on both stators as 0V/0v...-100V/+100V or 0V/0V...-90V/+110V shoudn't affect the result!?
In both cases is E = 200V / distance_of_the_stators.
4) Are there newer Versions or bugfixes of the original version? What about the schematic in this thread?
post #2 of 7

If you already know of the other place - it has lots of diy ES amp activity – search for kgsshv – look in that project thread (and T2) for some HV Q parts availability, substitutes


I think the IXTP01N100 is good, can be used as floating ccs, your AM6 – in practice a gate series damping R of ~ 50 Ohms is needed to prevent oscillation in the MHz



PS regulation is a biggy for Gilmore and fans – but the numbers aren't that convincing to this EE type

as a balanced circuit from input to output the BH amp circuit has high PSRR


for the amount of capacitance they use a CRC could drop supply ripple below 50 mV


even 1 V differential across the headphones would be inaudible because the sensitivity is so low

1 V on 100dB/100Vrms Stax gives 0 dB SPL, which is at the threshold for human hearing in an anechoic chamber

and you'd never get as much differential on the outputs as the PS ripple due to the balanced circuitry and feedback loop gain adds more PSRR


if you keep the regulation the power supply zener string bias doesn't need to be a active ccs – but you also don't really need the kV Q in that circuit

at worst add a protective 150 V zener from collector to raw rail to protect a 150-200 V pnp – lots more available


I would upgrade to a regulated diaphragm bias supply – at the BH lower supply V you need a V doubler and then could steal the kgsshv bias circuit



search “cathode stripping” - not really a problem below multiple kV


heater filament wiring AC current magnetic field leakage/coupling to other parts of the circuit are likely bigger potential problems than AC heater drive inside the tube with the cascode topology – the tubes simply are conducting whatever current the 2SJ79 are pulling


Linear Systems is currently making LSK389, and a new LSK489


you can hear music with just one stator driven but there is added distortion as the diaphragm-stator spacing changes the force/V changes

by using balanced drive from both sides the force is more linear with diaphragm displacement


Edited by jcx - 7/28/13 at 9:38am
post #3 of 7

It would be best to move this over to the other site (in case you wonder why we don't mention it by name, it's simply not allowed here) but either of the IXYS parts would work.  Now Mouser isn't allowed to sell to Europe anything IXYS as Arrow is the distributor so you'd have to get it from them. 


For the PSU, we do prefer regulation but a simple pass transistor and a zener string would also work just fine. 

post #4 of 7

there is nothing wrong with the T2 current source either if you can get the parts.

But it is a bit more complicated.


lil-knight manages to get the ixys parts from mouser directly, how he gets the

parts over there I do not know. The parts are actually made in malasia.

post #5 of 7
post #6 of 7
Thread Starter 
Thx all, you mean places 1) with doItYourself in the address or 2) with "h" and a "-" or 3) the one with the original Blue Hawaii Document?
I end up in the following possibilities (the switch at AM4 symbolizes the "mu follower Output"). 


If the only advantage of AM4 is less phase shift in higher audible freq's I'd keep it in mind but wouldn't realize it now. 
My experience with FIR filter and phase shift is roughly "one can hear it, it changes the soundstage but not really
the general impression to good or bad."
From the PSRR viewpoint I'd start with the concept: "care about higher harmonics, PSRR can start to decrease at a cut-off
in the 1 kHz region". So what about a plain and simple 6dB/oct low-pass with R*C*2pi=1Hz like that? (Maybe R2 is a bit verry low,
but in the newer schematic I found a 12 mA CCS, so I hope for stereo no much more than 50 mA are needed.)


Simulation (also with R1=4k) gives ripple in the 2 mV range, the higher harmonics in the uV range, probably lower than other distortions.
So, what about the http://gilmore.chem.northwestern.edu/BlueHawaii-T2-Servo-v1-2.zip Version mapped to the original one (i.e. without
the servo stuff for the beginning)?
post #7 of 7

pretty sure they mean "h" and "a" ... that's more or less where they live ... wink.gif

New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Need help in details building Kevin Gilmore's Blue Hawaii Amp