USB Type-C PD Lab Power supply

[amitchell]

I have been working with USB type-C Power Delivery lately and have been thinking about how it may be a useful source of power for a constant voltage/current lab power supply.

100W - 20V 5A is a pretty decent source to draw from.

Any thoughts?

[amitchell]

Yeah I suppose or any other available contract, how often do you need more than 20V? Could also Buck-Boost it.

I think it gets neat when you can power it from your PC/laptop or your usb-c enabled peripheral, monitor etc.

[amspire]

If you intend having the option of running the supply off your computer or anything else that is earthed, it would be a good idea to make sure there us full isolation between the USB C rails and the power supply output. Power supplies get abused a fair bit.

It is much easier controlling switching noise in a non isolated converter then an isolated one, so the easiest way to achieve this may be to have a simple push-pull continuously running fixed output isolating converter, followed by the variable output voltage switching regulator for the power supply.

Even if you are designing this for a dedicated USB C mains supply, it would be very handy to be able to run this off your computer, even if it is at a much lower power.

[krho]

AFAIK the cables are specified for xxV@3A for everything else you need special cables. So max you can get is 20V*3A

[David Hess]

If you intend having the option of running the supply off your computer or anything else that is earthed, it would be a good idea to make sure there us full isolation between the USB C rails and the power supply output. Power supplies get abused a fair bit.

I was going to say this but you beat me to it.  A non-isolated design will work if no other line connected equipment is used but USB was not designed to operate between devices on separate grounds and it was not designed to power devices on the same ground; in both cases a ground loop between ground and negative is created.

It is much easier controlling switching noise in a non isolated converter then an isolated one, so the easiest way to achieve this may be to have a simple push-pull continuously running fixed output isolating converter, followed by the variable output voltage switching regulator for the power supply.

In this case I doubt using an inverter for galvanic isolation followed by a switching regulator will help.  Using one of the transformer isolated topologies should work well enough and a single stage will be more efficient.

If I wanted low noise, then I would use an inverter followed by a linear regulator.

[amspire]

It is much easier controlling switching noise in a non isolated converter then an isolated one, so the easiest way to achieve this may be to have a simple push-pull continuously running fixed output isolating converter, followed by the variable output voltage switching regulator for the power supply.

In this case I doubt using an inverter for galvanic isolation followed by a switching regulator will help.  Using one of the transformer isolated topologies should work well enough and a single stage will be more efficient.

If I wanted low noise, then I would use an inverter followed by a linear regulator.

Theoretically, what you say is true. When it comes down to the actual design, it is just much harder to achieve a wide ranging transformer isolated variable output converter then a two stage solution. The USB-C voltage can be 5v to 20V. Whenever you have a wide ranging input output, a single stage transformer-regulator that doesn't product massive ringing is extremely challenging. I am talking about noise that is really hard to contain to the supply. Designing the transformer will be hard.

If you have a simple continuously running isolator that converts 5V to 15V or 20V to 60V, that is a simple, trouble free design. The transformer is very simple.

Follow it with a buck converter that can output 0-12V from a  5V USB-C source or 0-50V from a 20V source is also very straightforward with either an off the shelf inductor or something like a simple toroid core.

Go for the challenging design if you like.

[amitchell]

AFAIK the cables are specified for xxV@3A for everything else you need special cables. So max you can get is 20V*3A

An appropriate EMCA (Electronically Marked Cable Assembly) is needed for the full 100W, that is not a deal breaker. I would certainly purchase a high quality appropriate cable. The power contracts negotiated between source and sink (UFP and DFP) take into account the cables power capability.

If you intend having the option of running the supply off your computer or anything else that is earthed, it would be a good idea to make sure there us full isolation between the USB C rails and the power supply output. Power supplies get abused a fair bit.

Yes isolation would be smart. CC1 CC2 are needed as well as DP/DM assuming you would like to upgrade firmware over that pair.


When negotiating the power contract the device cannot present more than 10uF to VBUS and when the contract is negotiated it cannot present more than 100uF to VBUS. This will likely impact what topology can be used here.

[amitchell]

The table on the right is what the newer standard allows for.

[RGB255_0_0]

Even if you somehow manage to achieve a variable output, not all type-c supplies are the same; and perhaps that is the same when drawing from the devices mentioned in the OP. See pic:

[garboui]

This chart is likely a result of versions of the standard and what operating modes were chosen to be supported by the manufacturer.

I have been thinking of a similar project that last couple months. One idea is to have the PS support all modes then present the user with the highest power option available (or selectable); the range and op is then limited to this.

[amitchell]

This chart is likely a result of versions of the standard and what operating modes were chosen to be supported by the manufacturer.

I have been thinking of a similar project that last couple months. One idea is to have the PS support all modes then present the user with the highest power option available (or selectable); the range and op is then limited to this.

The firmware of the USB-C controller IC should be able to do this automatically. I believe the Cypress CCG3 that I have been using is capable of this.

[RGB255_0_0]

This chart is likely a result of versions of the standard and what operating modes were chosen to be supported by the manufacturer.

Quite likely. Also possibly due to manufacturers wanting to protect themselves from poor quality cables after it made the rounds last year. Perhaps some companies (HP as in the Spectre) "broke" the spec with their own.

One idea is to have the PS support all modes then present the user with the highest power option available (or selectable); the range and op is then limited to this.

I think that's your only possibility. Just don't expect it to work with every USB-C power source, as shown in the table, or Murphy will get you.