By looking at the excellent
teardown of the Mastech HY3005F by jamesp15 and his last two pictures with all the transistors attached to the beefy heatsink and mounted with the silicone TO-3 insulators, I would like to tell my story with these... (together with a small teardown)
Since 2002 I own an Instrutherm FA3005 (Mastech HY3005 clone manufactured in Brazil, 0-30V 0-5A single output) and the build is overall very good.
Frontal view.-fix-and-teardown/?action=dlattach;attach=244993)
All images by
rsjsouza at 2012-08-07
The first thing that caught my attention is that there is heat shrink tubing all around and several wires are protected by woven fabric insulators when crossing high power components - although some 5W resistors touch some wires when the main board is completely in place (not a big fan).
Side view. Cables with woven fabric insulation.-fix-and-teardown/?action=dlattach;attach=244995)
Transformer output cables with heat shrink.-fix-and-teardown/?action=dlattach;attach=244997)
Voltage selector switch, fuse and mains connector.-fix-and-teardown/?action=dlattach;attach=244999)
Current sense resistor and power switch.-fix-and-teardown/?action=dlattach;attach=245001)
Looking at the solder side of the main board one can tell it is clearly hand assembled, but I couldn't spot bad or flimsy solder joints. Also, multiple 3/4 turn pots in series for fine trimming are used instead of the multiturn, low PPM ones. But all in all, great power supply. There is a mix of electrolytic capacitors rated for 85°C (blues) and 105°C (blacks), good quality transistors and ICs (2N3055's from ST, other IC's from Hitachi and TI) and the voltage/current meters are die-on-board 7106's! Interesting is that in 2002 both the hot melt glue and the "fan-over-heatsink" frenzy weren't commonplace yet...
Main board solder side. I applied a silicone insulator that causes the strange shiny effect on the board. -fix-and-teardown/?action=dlattach;attach=245003)
Main board component side. -fix-and-teardown/?action=dlattach;attach=245005)
Detail of the relays, main rectifier diodes and 5V voltage regulator. Interesting to see that the board says HY3002D-3(5), which indicates it is a shared design with the HY3005.-fix-and-teardown/?action=dlattach;attach=245007)
Detail of the main filter capacitor, shunt resistors and the main ICs. -fix-and-teardown/?action=dlattach;attach=245009)
Detail of the voltage (left) and current (right) meters. The interesting detail are the ICL7106 die-on-board ICs.-fix-and-teardown/?action=dlattach;attach=245011)
The reason for the teardown was a long-existing problem that caused me some headache: the output voltage was tied to the mains voltage (the +/- terminals had 12/48Vac difference to the ground). Something similar to what Dave explains in
episode 279.
For a long time I thought this could be related to the use of an autotransformer (for several years I blamed the manufacturer for being
that cheap), but when I found the schematics in the web the other day I decided to investigate this further. There it does not show anything connected to the mains GND, although the multimeter read several kiloohms between the GND pin and the positive output. So it must be related to the build.
With the multimeter I started inspecting the +/- supply lines on the power supply until I found that the cathodes of D1/D2 (positive output of the bridge rectifier) were shorted to the chassis - this line is also connected directly to the collector of the 2N3055's mounted on the heatsink (V20/V21/V22), which is a critical spot due to temperature and complex mechanical assembly.
Schematics detail showing the devices shorted to the chassis.-fix-and-teardown/?action=dlattach;attach=245013)
I then unscrewed the entire heatsink from the chassis and the short between GND and the power line was gone - thus the heatsink mount, and more specifically the TO-3 insulation, was the problem. I loosened each 2N3055 screw one-by-one and measured with the ohmmeter to precisely locate it. The issue, obviously, was evidenced only in the last screw...
I had to de-solder all three 2N3055s, as they are mounted in a very nice PCB close to the heatsink, and removed all the insulation assembly. I then found that the silicone insulators showed round cuts around the edges of the transistor's emmitter/base pins, and the internal part of the heatsink revealed a sharp edge caused by the drill.
The three TO-3 insulators. Notice the rings around the pin holes. -fix-and-teardown/?action=dlattach;attach=245015)
Detail of the most critical insulator. Interesting to see the transistors were numbered, probably to match its characteristics.-fix-and-teardown/?action=dlattach;attach=245017)
Therefore, when the transistors were mounted on the heatsink, the sharp edges touched the TO-3 can, which shorted the collector of the transistor to the chassis of the power supply - which is connected to the mains GND.
I then had to decide if I would insulate the transistors from the heatsink or the heatsink from the chassis. I opted for the second alternative for the following reasons:
- I found in the local hardware store some thick (1/4'') nylon screws with high melt temperature. This still guarantees a great mechanical robustness in addition to the electrical insulation;
- The thermal contact between the transistors and the heatsink is greatly improved when compared to mica or silicone. These two app notes from TI/Burr Brown are good reading: check
here and
here.- The power supply sits most of the time in my lab bench, thus the mechanical robustness lost by using the nylon screws is not a critical factor - nor the eventual guy that would be tempted to carry it by the heatsink!
Therefore the final transistor assembly is shown below:
The final assembly without its cover.-fix-and-teardown/?action=dlattach;attach=245019)
PCB view of the TO-3 assembly. Normally this board would be completely inserted into the box. -fix-and-teardown/?action=dlattach;attach=245021)
Detailed view of the 2N3055 and the upper nylon screw. -fix-and-teardown/?action=dlattach;attach=245023)
Side view of the upper nylon screw. Notice the distance created between the heatsink and the chassis. -fix-and-teardown/?action=dlattach;attach=245025)
Side view of the TO-3 assembly. Although the heatsink lost physical contact with the chassis (which helped with heat dissipation), the air gap and the direct contact of the transistors greatly increased the thermal efficiency. -fix-and-teardown/?action=dlattach;attach=245027)
The morale is to always think of heatsinks as a critical point in the design... Even the smallest detail can cause great havoc (I lost a PC motherboard due to this issue, but that's a story for another time...)
The complete schematics are in the link below:
-fix-and-teardown/?action=dlattach;attach=245029)
Home
Help
Search
Login
Register
