2
\$\begingroup\$

I am using this as rectifying diode in full bridge configuration in one of my power supply designs. My questions are regarding heatsinking of the diodes. They are as follows:

  1. Do I need to use insulation to isolate the diode's thermal tab and the heatsink from each other? If yes, should I connect the heatsink to Ground?

  2. If insulation is not used, should the heatsink be connected to the rectifying diode's thermal tab terminal (cathode) through the PCB, through the heatsink's mounting screws?

The design specs are:

Input Voltage: 90V to 240V AC

Output Voltage: 55V DC

Output current: 7A.

[Edit]

The power supply has PFC+LLC configuration. The part specified above is used to rectify the output of the LLC stage in bridge rectifier configuration as shown below:

The anodes of the part have been connected together to use each part as a single diode

Thanks in advance.

\$\endgroup\$
  • 1
    \$\begingroup\$ Did you mean full-wave with 2 diodes and CT transformer, instead of full-bridge, which normally means 4 diodes? \$\endgroup\$ – Mattman944 Jul 8 at 15:28
  • \$\begingroup\$ What freq. and how do you intend to regulated Vdc? \$\endgroup\$ – Sunnyskyguy EE75 Jul 8 at 15:47
  • \$\begingroup\$ @Mattman944 the two anode terminals have been shorted to use the part as a single diode \$\endgroup\$ – mwaghmare Jul 9 at 10:09
  • \$\begingroup\$ @SunnyskyguyEE75 the operating frequency for the LLC stage is 70kHz to 350kHz. \$\endgroup\$ – mwaghmare Jul 9 at 10:13
1
\$\begingroup\$

Do I need to use insulation to isolate the diode's thermal tab and the heatsink from each other? If yes, should I connect the heatsink to Ground?

If insulation is not used, should the heatsink be connected to the rectifying diode's thermal tab terminal (cathode) through the PCB, through the heatsink's mounting screws?

Summary:

Heat sink isolation of the diodes is required OR individual heatsinks that "float" could be used, but this is less common.

There are numerous insulating thermal mounting systems available for this package. Thermally conductive electrically insulating rubber pads are easy to use and well priced.

Heatsink grounding is usual but may depend on application.

________________________________________

Detail:

As the packages have the Cathode connected to tab, in the circuit shown D25 & D26 could share an electrical connection to a common heatsink - but it would then be at Vout_positive = 55 VDC. This is not a "problem" if sufficient allowance is made for this, but it would be more usual to ground the heatsink and to use low thermal resistance insulating "washers" between the packages and heatsink.

D23 & D24 Cathodes and therefore package tabs need to be electrically isolated from any other electrical connection. If they used separate heatsinks they could be electrically connected to the package tabs but this would be 'unwise'. As above, use of a suitable insulating washer is recommended.

If insulating washers are used on all diodes then a shared heatsink can be used.


Dissipation:

Data sheet says Vf forward voltage at 8A is 0.85V typical.
At 7A then dissipation = I x V = 7 x 0.85 ~= 6 Watts.

Use of the two diodes per package in parallel will reduce the net wattage dissipated - but see below re peak currents.

Thermal resistances (page 2) are:
Rth_JA = 50 C/W = junction to air .
Ryh-JC = 1.25 C/W (both leads) = junction to case (leads)

SO at 6 Watts dissipation with no heatsink temperature rise is
Pd x Rth_JA = 6W x 50 C/W = 300 degrees C temperature rise.
It may not quite glow in the dark, but it also will not last very long at all.

A 10 C/W heatsink will give you 10 x 6 = 60C rise.
That's for one diode.

If you have an AC 4 diode bridge then duty cycle is 50% or less and heating is half or less what is expected.

Dissipation can easily be much higher than expected if significant output capacitance and consequently low ripple voltage occurs. When feeding rectified DC to a capacitor that is large enough that the voltage does not vary significantly across a power cycle means that the diodes will conduct only near the waverform peaks, and current may be in bursts of much (or very much) higher than average current and you need to calculate accordingly. Use of a series "spreading resistor" to increase the conduction angle of the diodes may well be in order.

As a "bonus", rectifiers with high current peaks can generate 'useful' amounts of EMI.


References:

Each thermal material has a specification that can be used as part of the design calculation. Most materials designed for this application have a low enough degrees/Watt thermal resistance that while they should be accounted for , they are not major components in design decisions except in extremely high power applications. eg

this 3M thermal tape has a 2 mil / 0.05mm !!! version rated at 0.35C/in^2/Watt and 1500 V rated. $US10c/in^2
Range of Digikey sourced products here
and more here
and this magic material* achieves 0.013 degrees C / Watt per inch^2 - but does not mention electrical resistance - but is "probably" a usable insulator. Probably. At about $US0.25 per square inch it seems reasonable value for what it achieves.

\$\endgroup\$
  • 1
    \$\begingroup\$ "Thermally conductive electrically insulating rubber pads..." They never work as well as mica insulators and thermal paste, but they're a lot less messy. I used to work at a place where there was a running battle between engineering, who wanted mica and goo, and manufacturing, who wanted to use SilPads. If you're going to use pre-made pads, be sure to take them into account in your thermal design. \$\endgroup\$ – TimWescott Jul 8 at 14:58
  • \$\begingroup\$ @TimWescott Yes (but :-) ). Each thermal material has a specification (as I know you know) that can be used as part of the design calculation. Most materials designed for this application have a low enough degrees/Watt thermal resistance that while they should be accounted for , they are not major components in design decisions except in extremely high power applications. eg this 3M thermal tape has a 2 mil / 0.05mm !!! version rated at 0.35C/in^2/Watt and 1500 V rated. ... $US10c/in^2 \$\endgroup\$ – Russell McMahon Jul 9 at 7:38
  • \$\begingroup\$ ... Range of Digikey sourced products here \$\endgroup\$ – Russell McMahon Jul 9 at 7:40
  • \$\begingroup\$ ... and more here \$\endgroup\$ – Russell McMahon Jul 9 at 12:26
  • \$\begingroup\$ ... and this magic material* achieves 0.013 degrees C / Watt per inch^2 - but does not mention electrical resistance (but is "probably" a usable insulator. Probably. At about $US0.25 per inch^2 it looks 'reasonable' value for money. Prices here \$\endgroup\$ – Russell McMahon Jul 9 at 12:29
0
\$\begingroup\$

You would need three of them to get a full bridge.

There is no electrical necessity to isolate the heatsink from all three dual diodes but you do need to isolate at least two since the common cathodes/heatsinks must be connected to three different places.

There may be safety reasons to isolate all three and ground the heatsink or leave it floating within an insulated housing but there is not enough information to be sure about that.

You definitely need to make sure nobody can come into contact with it which powered if it is not isolated OR if it is floating.

\$\endgroup\$
  • \$\begingroup\$ You only need one of them if the supply is a center tapped winding, which is much more likely.. \$\endgroup\$ – Jack Creasey Jul 8 at 16:29
  • \$\begingroup\$ 4 parts have been used to form a full-bridge as I have edited in the original question. The two anode terminals have been connected together to form on diode \$\endgroup\$ – mwaghmare Jul 9 at 10:30
  • \$\begingroup\$ @JackCreasey no the transformer center tapped winding is not used. \$\endgroup\$ – mwaghmare Jul 9 at 10:31
0
\$\begingroup\$

You won’t get 50Vdc out of this design as provided since the Vac peak voltage varies from 130 to 340Vp.

These all use active tabs that must be insulated with mica and silver oxide grease for electrical insulation, then you can earth ground the heatsink.

But worse, you have huge surge currents into a low ESR cap and huge dynamic range of DC for input tolerance and unspecified load tolerance.

Therefore, I suggest you do not specify the diodes but rather specify to the dynamic Load impedance for your wide ranging Vin. An active PFC design is recommended for all power supplies >100W and is now mandatory for commercial designs.

If you still want a brute force rectifier with some regulated DC then I suggest the method perfected by Lambda and HP in the ‘70’s using a phase controlled bridge for the pre-regulator to get just above the 50V spec. driving these can be done with simple signal transformers or miniature wire-wound iron wire cores and logic levels to trigger the phase of the triac full bridge with DC servo feedback.

E.g. https://media.digikey.com/pdf/Data%20Sheets/ST%20Microelectronics%20PDFS/ACST10.pdf for a sample Triac.

Research Triac or SCR full bridge designs for line voltage regulators AC-DC and choose a series choke to resonate in series with load cap to store the dI/dt current pulses to smoothen the voltage output.

A medically approved commercial supply with active PFC and excellent reliability with 90 to 95% efficiency at 400W, here is an example of one that costs only $0.65/W

https://www.mouser.ca/ProductDetail/Murata-Power-Solutions/MVAC400-48AF?qs=sGAEpiMZZMsPs3th5F8koDM6RoqAUmtevI1toRmi7jY%3D. You can start from this as a comparison to your output specs. Then search for others.

\$\endgroup\$
  • \$\begingroup\$ Yes, the power supply uses a PFC+LLC configuration and the part specified in question is used to rectify the LLC stage output. \$\endgroup\$ – mwaghmare Jul 9 at 10:34
  • \$\begingroup\$ But how is it regulated? \$\endgroup\$ – Sunnyskyguy EE75 Jul 9 at 14:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.