Since Icom's latest improvements to the IC151 and driver circuits, I've come to realize that additional heatsinks may be overkill, at least in regards to IC151. It has become clear that ESD (electrostatic discharge) was the primary cause of IC151 failure. And the added diodes on IC151 and the HRX-line seems to have greatly improved reliability. Also, the reduction of idling current has improved reliability of the driver circuit (DRV board).
That being said, adding heatsinks certainly can't hurt. The predriver (Q1) and driver FET's are surface-mount components that rely primarily on their solder connections to transfer heat. The stress of hot/cold cycling over time can cause the solder bond to weaken, eventually causing intermittent operation or complete failure.
For better cooling, manufacturer's often place thermal-pad material between hot-running surface-mount components and the metal chassis, in lieu of a dedicated heatsink. And sometimes they create a "heatsink sandwich", like Kenwood did in the TM-271. Its driver FET is bonded to the chassis under it, and there's a heatsink above it, with a thermal-pad in between. This is where I got the idea for the heatsink/thermal-pad modification described below.
But I wouldn't bother adding a heatsink unless there is actually a failure. I would only add a heatsink if replacing or re-soldering component(s) becomes necessary. If it ain't broke, don't fix it!
Driver Idling Current:
Some folks have set the driver idling current lower than the latest factory-recommended setting of 1.5 amps. I don't recommend doing so, as this can increase IMD (inter-modulation distortion). Click here to see spectrum analysis of an IC-746PRO with idling current set to 1.5 amps, compared to 0.5 amps.
Some IC-746PRO's were manufactured with the wrong value resistor in the predriver circuit. R3 on the PA board should be 10 ohms, but some boards had a 4.7 ohm resistor instead. This can cause failure of the predriver FET (Q1).
The IC-746PRO is a great rig. But it has a few reliability issues, especially with the earlier-production units.
This article addresses four problem areas:
NOTE: The following information is based on my experiences and observations. I have not consulted with anyone from Icom. And I do not recommend doing any modifications to a properly functioning radio. If it ain't broke, don't fix it!
The first issue is SMD-chip IC151 on the RF Unit. This chip can fail for one or two reasons: electro-static discharge (ESD), or heat.
ESD: Icom released a service bulletin (#920), which recommends adding two PIN diodes to the HRX-line. This can help prevent ESD-related failure.
Heat: It is a commonly-held opinion that this chip runs too hot. It seems Icom did not design the PC board with enough ground-plane to cool the chip properly. Adding a heatsink on top of the chip, soldered to the ground pins, can alleviate the problem. I've found that a 2AG-style fuse-clip fits perfectly onto the chip, and serves as a decent heatsink.
Later-production IC-746PRO's have a couple of diodes soldered on top of IC151 (see the picture below). In this case, soldering a heatsink on top is not practical. I've drawn a diagram of the factory modification. The SMD diode is an MA77. There are no visible markings on the other diode. I assume this upgrade is to prevent damage from ESD (as if service bulletin #920 wasn't enough).
In the latest-production units, Icom has revised the RF Unit board to accommodate three extra SMD diodes in the IC151 circuit (for additional ESD-protection). Its basically the same as the circuit described above, except it uses the smaller version of the MA77 diode (MA2S077). Two of these MA2S077 diodes are in series, in place of the large amber diode in the previous circuit. The other MA2S077 is used in place of the MA77 in the previous circuit. A picture and schematic are shown here:
The second issue is the driver circuit (DRV Board). In the earlier production units, the two FET's (2SK2975) run quite hot. Mostly because the factory set the idling current (bias) too high. In addition, the FET's are surface-mount types, which dissipate their heat through their source connections. But the manufacturer's heatsink doesn't seem to do a very good job. The excessive heat eventually causes the solder connections to become intermittent, which causes fluctuations of the transmitter RF output. This affects all bands and all modes, and it is most noticeable by observing the ALC meter.
If the FET's are still good (and they usually are), they can be re-soldered. But there's a trick to it. The trick is to melt the solder on all three of an FET's strips at same time, while holding the FET firmly against the mounting plate, so it doesn't slide out of place. If you've re-soldered without the FET becoming loose, you didn't get it hot enough, and the joints will fail again. Unless you have the proper soldering equipment and skills, its probably best to replace the entire DRV Board.
After the FET's are re-soldered or replaced, the the PA idling current adjustment (R11) should be set to 1.5-amps (instead of 2.5-amps, as stated in the service manual).
You may notice the factory did not apply heatsink compound under the DRV Board. You can add some, but I don't think it makes enough difference. The metal is machined so smoothly that heat transfer isn't greatly improved by adding heatsink compound.
To provide additional cooling, a heatsink can be added on top of the DRV Board. This is done by inserting a 1mm-thick thermal pad between the FET's and the new heatsink, so the transfer of heat is accomplished without putting any mechanical pressure on the components.
The original mounting screws are replaced with longer ones (of the same thread-type) to accommodate the new heatsink. Two metal spacers are inserted to keep the heatsink at the proper distance above the FET's (just enough for the thermal pad to be slightly compressed). The optimal screw length is 13mm to 14mm, not including the head. The spacers should be metal, 2mm thick.
Here is a cross-sectional view of the modification:
Here are "before" and "after" pictures of the modification:
The heatsink and thermal-pad material are available from Digi-Key. Here are the part numbers:
Thermal Pad: BER162-ND
NOTE: Re-soldering the FET's is no easy task. It requires proper tools and skills. The trick is to melt the solder on all three of an FET's strips at same time, while holding the FET firmly against the mounting plate, so it doesn't slide out of place. If you've re-soldered without the FET becoming loose, you didn't get it hot enough, and the joints will fail again. To get an idea of how the FET's are soldered, here are some photos, courtesy of ZD8I:
Personally, I prefer not to remove the DRV Board. Instead, I remove the entire PA Unit. This may be more labor-intensive, but it eliminates the risk of damaging the traces of the DRV Board, as it requires considerable heat to unsolder its mounting pins. Removing the PA board only requires unsoldering the VHF SO-239 connector and two of the Varistor Boards.
In the latest-production units, Icom has made significant changes to the PA Unit. Retrofitting an older unit requires extensive modifications. Here are the changes made to the latest PA Units:
Another common problem is failure of the HRX-muting circuit. This is usually caused by RF overload, high SWR, or lightning. If this circuit fails, the transmitter will tend to oscillate intermittently. This is most noticeable on the higher HF bands (6, 10, or 12 meters). 2-meters will not be not affected. The symptoms are: unusually high current demand from the power supply, fluctuating RF output, and/or erratic antenna-tuner operation.
To get an idea if the HRX-muting circuit is at fault, turn the receive attenuator (ATT) on. If the problem goes away when the ATT is on, the HRX-muting circuit is probably at fault. This test works because the attenuator interrupts the path of oscillation.
Suspect one or more of the following components on the CTRL Unit:
D21, D22 (MMBV3700LT1)
Note: To check the DC voltages at Q25, transmit on SSB with the mic-gain at minimum. The base voltage should be 0-volts RX, and 0.7-volts TX. The collector voltage should be 13.8-volts RX, and 0-volts TX.
In the latest-production units, Icom has made changes to the CTRL Unit. Here are the changes made to the latest CTRL Units:
This is less common, but the predriver FET (Q1 on PA Unit) will sometimes become intermittent or fail completely. This is a 2SK2973, and it runs quite hot. When I replace the FET, I solder a small heatsink to its source tab. A 2AG-style fuse-clip fits perfectly by cutting off one of its legs, then soldering the remaining leg to the FET tab. See the pictures below:
NOTE: If you don't have the skills or equipment to work with static-sensitive surface-mount components, please leave it to a qualified technician.
DISCLAIMER: Attempting these modifications on your own equipment is at your own risk. If you have any doubts, don't try it.
The heatsink modifications were not endorsed by Icom.
Mike Nadeau - N1EQ