Blue hawaii build notes for the kgsshvpssicfetdual2new PSU
The kgsshvpssicfetdual2new is based on version 1.6 of the kgsshv psu single boards and does NOT have the cpc1117N or resistor required for simple B+,B- delayed start. (it’s still possible to implement delayed start but it requires a relay(s) which the cpc1117N avoids). In all other ways its similar to the v1.7 single boards. The single boards put the b+ on one board and the b- on another and split the -+15V and bias between them.
The kgsshv ps dual has all the dc voltages required to power the blue hawaii and apart from the notes above is topologically and componentry identical to the single psu boards. The notes bellow applies to BOTH the single and the dual psu boards.
Construction is straightforward other than
1. the silkscreen shows 15-0-15VAC input for the + and 15VDC outputs, Since the -+15V output is a simple DC supply with a bridge rectifier and voltage regulators 15-0-15VAC is NOT sufficient to provide -+15V output... I have verified this using a variac. There is about 0.6V drop across each diode in the bridge *plus* the voltage regs need between 1V and 1.5V more input DC than their output in order to regulate properly. so 18-0-18AC is actually required to get a properly regulated -+15VDC output... Any more VAC input will be converted into heat by the voltage regs. However, with 18VAC-0-18VAC centre tapped input the voltage regs run very cool with the small heatsink they are bolted to.
2. the blue hawaii amp boards -400VDC lines draw more current than the +400V lines. If you connect a single channel up to the psu board for testing there will be no problems. Connecting two channels results in the -400V line on the psu going into current limit and the voltage dropping to (im my case) about -175VDC. The fix is fortunately easy. The 5.1ohm 3W or 5W resistors are the current sense resistors for the over current protection. On the -400V psu side either replace the 5.1ohm resistor with about 2.6ohm or similar wattage and type or parallel another identical 5.1ohm around the existing one. The +400V does not need modification. the lower the resistors value the more current supplied before the current limit cicruit activates. I don’t have anyway to variable load a line that has such a high voltage but I would guesstimate that the 5.1ohm resistor current limits at about 100mA since two amp boards +400V together draw around 81mA and there is no problem with 5.1ohm and 2 -400V boards draw between them around 129mA and that causes current limiting with 5.1ohm sense resistors.
3 make sure none of the metals tabs on the transistors are shorted to the heatsink/L bracket that mounts to the heatsink. Use electrically insulated spacers between the back of the transistors with metal tabs and the L bracket/heatsink AND use a nonconductive gromet to insulate the bolt from the transistor. The psu transistors do not get that hot with a reasonable size heatsink attached.
4. if you are building in a u2 high case the largest main psu caps you can put in are Kemet 500V 470uF long life at 65mm tall. Beware they also do an 80mm tall 470uF cap - this will NOT fit in 2u. Similarly 680uF caps at 500V are 80mm high and will not fit.
5. if you want to check the bias is 580V use the bias test point. Dont measure at the bias screw terminal. Why, typical multimeters have a 10M ohm input impedance and assume whatever they are measuring across is much less than 10Mohm... this is NOT the case for the bias line which has a 4.7Mohm resistor. result the multimeter will NOT give the correct voltage reading between the ground screw and the bias screw. Measure from the bias pad nearby to the ground screw. Also, don’t measure the bias with headphones attached.
6. there is only one screw terminal for +400V, -400V, +15V and -15V so you will need to connect two wires to each screw terminal. This limits the gauge of the wire. Also make sure you are screwed the terminal down well, its easy to have one wire firmly affixed and the other one lose....
7. to adjust the B+ and B- lines the two 390K resistors (R8 and R9) in series with the 20K resistor (R7) just after the 1N007 diode that have the 0.047uF cap across them need to be changed. Approximately (for my psu) I got the following
442K for 450VDC (note requires a transformer with 360VAC output to have enough voltage input to get proper regulation)
390K for 404VDC (transformer 330VAC or more output)
365K for 375VDC (transformer 310VAC or more output)
the approximate equation for me was the total of the 2 resistors in ohms/(20,000-700)*10
(the 20,000 is the resistor in series with the resistors you are changing, the 10 is the output of the voltage reference LT1021-10), the 700 is the fudge factor to make the numbers accurate for me... (variation in voltage reference, variation in the 20K resistor etc etc...)
8. To test for proper regulation connect a volt meter to the nominally +-400V output and a ground screw terminal and either power the transformer from a variac OR if your variac can go high enough you could go from the variac output into the 300VAC input of the psu. There is no need to have amp boards connected at all and its safer if you dont. Slowly increase the AC and you should see the DC output rise. At some point increasing the variac by some volts will result in very small output increases (in the order of few milli volts) now the psu is regulating. Measure the 330VAC input of the psu and that’s the minimum transformer output voltage you need for regulation. Both the + and - 400V sides of the psu should have similarly behaviour and similar minimum VAC input requirements since they are almost identical in topology and are identical in components.
9. I would recommend initially testing the psu board without amp boards connected in case voltages are way off. However, with no load on the psu the b+ and b- lines, they will take some minutes to fall to 0v and with no headphones the bias line will stay high for a long time. Fortunately the bias line can be discharged just by connecting a multimeter to it and ground and set to DC volts. Even a meter with 10Mohm input impedance will drain the bias line in a few minutes. the -+15V lines will be drained by the voltage regulators.
10 I would recommend a variac for initial testing it allows you to bring voltages up slowly and make sure that voltages are going up in symmetry (-+400V, -+15V) and becoming stable.
11. If the outputs look good with no amp boards attached, connect one board and repeat the tests. Better to put in channel in danger than both... especially if the amp boards have not been tested.
12. if test with the other amp board only. You don’t want to blame the psu if one amp board only is miss behaving...
13. test with both amp boards. Remember if the -400V line goes down in voltage significantly only with 2 amp boards attached and is fine with one you probably have a current limit problem and did not implement item 2 here!
14 don’t plug in headphones unless 1. you have a spare pair you don’t mind frying. 2. you have tested voltages, dc offsets are nulled and you are happy. ideally look at the outputs on a scope and put some sine waves in....
15 Before you plug in anything visually inspect for solder bridges... I had a partial bridge it only started conducting when the input AC to the psu went above about 26VAC... thank you variac...
16 use probes with insulation, there is lots of high voltages don’t short out things with you probes....
17 Before you power up anything double check the wiring the - and + on the silkscreen are quite small and you dont want +400V going to a place expecting -400V....
18 temperatures are low, with the psu attached at a 200mm by 80mm heatsink with 40mm fins, (both amp boards running constant current 20mA and -+400V B-+), the centre of the case of the +400V side transistors are about 40C, the -400V transistors are around 3C higher at around 43C. The cases on the C2M1000170D transistors take some time to become as hot as the mounting screws. The 10M90S cases reach the mounting screw temperatures far faster.
19 I know the board has been made as small as possible and around the screw terminals its rather crowded. But it would be nice if there was sepertate screw terminals for the -+400V and -+15V for each channel, this would allow for the use of thicker wires and reduce the possibility of a wire comming out. (I have already had the situation where I though both wires where screwed in tightly to a single terminal and then one came out)
Hope this is useful