Interview with S.Carlsen, commitee member for USB Power Delivery standardizationApril 3, 2013
After my recent posts (USB-PD Intro, PHY, Protocol Layer) about USB Power Delivery, I had the chance to come in touch with great people and I started with them interesting and promising discussions. In the early phases of any new emerging technology it’s always important to build a wide network of interest, competences and contributions.
In this sense, it is a pleasure to introduce Sten Carlsen, a member of the standardization committee of the USB-PD. Having covered this active role, Sten can help to better understand the new standard and read it out of the specification. Below the interview I had with him.
Q1 Sten, can you introduce yourself and explain your role inside the USB-PD standardization committee?
A1 At the beginning of the work on the specification, I was employed by Nokia/Copenhagen. I had been working for Nokia about 15 years at the time doing many different things, I have been involved in the USB charging work in various ways since spring 2005. I set the system description for our first generation of chargers, including the USB-OTG power function. The OTG needs the phone to generate 5V to supply the memory stick or what is connected, we used the same components to do both ways to save cost. The starting point for the work was that this communication should be orthogonal to the normal USB-communication. In the first meeting of the committee I had prepared my thoughts about the communications system and I was asked to head that work group. This was in mid 2010 but there had been some preliminary work done by a couple of guys, one of them my colleague from UK, so I knew something about it before it really started.
Q2 How much did take the committee to come out with the first release of the standard? Do you have any anecdote to share with us?
A2 It took about 2 years from the real start of the work until we released the first version. It was a very good example of team work, no sign of politics involved. I really have to praise the guys from the various companies for their commitment to getting a result out and for working together.
Q3 The specifications have been released on Aug. 2012. A complex ecosystem should be built to make this technology flying (USB-PD compliant cables, connectors and so on). So far, no product in the market or even announced. Do you have an idea why?
A3 As we have seen the specification is really different from previous USB specifications so designs need to be done and ICs finished before the next level of designs can be done. I do know that IP designs do exist and I know that products are under way. One announcement I know about was at the CES where SMSC announced they will soon have ICs.
Regarding cables and connectors, vendors are aware of the need and working. There are no news I am at liberty to disclose at the moment.
Q4 Can you figure out what will be the killer application? And how this standard will change our life?
A4 One clear design goal was to get rid of the many wall warts and replace that with just the USB cable.
One application I expect to be popular is the one where you have a laptop/xPad that you bring around and when you get to your work space you plug in the USB cable from the monitor and that single cable will provide video to the large display, connection to the mouse, keyboard, touch-“thing”, large hard drive for backup, network and in return it will charge your device. All of this has one power cord only – for the display. Your phone will obviously also charge from the display.
Q5 From a technical point of view, What the most difficult choice in the specifications?
A5 We spend much time on two items, the first one was the cable and connector question. Would we allow higher voltages on cables with legacy plugs? The current could be fine but the killer issue was the fact that cables with two A-connectors exist in volumes. Imagine that you plug such a cable into a device and a PC, then they negotiate for 20V – the second A-connector can then be plugged into an unsuspecting legacy port that may not survive getting 20V fed back into the normally 5V output. We spend a lot of time trying to detect that situation but had to give up.
The second item is the profiles, what should they look like? What would make sense? to devices? to people? to your mother? In the end we looked for economic balance i.e. making a power supply for one part of a profile has a cost and adding the other parts of the profile should not add further cost, well not more than needed.
Q6 The PHY is basically a powerline modem over USB. Can you elaborate more about and explain why they came out with this specifications (FSK, 23.2MHz carrier, 300Kbps)?
A6 There is an interesting development going before the final numbers. The first choice was to look at baseband signalling but at 5A, that is not really nice. Next we looked at using a carrier system, first thought was 1MHz, but as the SMPS at the cable end runs at up to 3MHz at the moment we needed to move above the harmonics. I know 6MHz SMPS have been considered, 3rd harmonic from them will be at 18MHz. The other thing is that for 20MHz a 5m cable is a half wavelength and there are some transmission line effects to care about. In the end 23.2MHz was agreed upon based on the various Ham radio frequencies in order to disturb as little as possible.
The modulation was chosen to be as simple to generate and receive as possible for cost reasons. Let me suggest just two examples of modulation:
1 – use an analog PLL and put the modulation on the frequency control pin with proper scaling and ranges.
2 – Should you have a 480MHz signal (HiSpeed) available, you can use a changeable divider and get a “1” by dividing by 20 and a “0” by dividing by 21. These frequencies are not spot on but clearly within limits.
Cost was always an issue in every discussion.
Q7 The power manager, able to deliver up to 100W and implement (optionally) 6 power profiles, is going to be a crucial and challenging block in a USB-PD device? Did you discuss implementation details? Did you make decision to enable a tighter level of integration with a USB-PD PHY?
A7 We did make some demo units early on in the process, these were equipped with both high voltage (20V) and higher current, limited by existing cables. Given that you do a careful design, there should be no problem, we saw no issue with the power and signal parts interfering, we did see some tuning issues as one of the demo units was based on a HC7046 and a breadboard with wires sticking out all places.
You need to be sure to look at the 1µH inductor in regards to the power supply. We have been running 3.5” disk drives off the demo boards with no issues.
Q8 The protocol permits to negotiate different power profiles, involving relatively high voltages especially for battery operated devices. What if the source is affected by a virus or malware? Can the source deliberately decide to rise the voltage and damage the sink, not ready to handle higher voltages? Do you think the protocol is safe enough to avoid such conditions?
A8 You can essentially negotiate any voltage and current within the limits of your cable. Whether there is a safety issue here depends very much on the actual implementation.
One of the demo systems did all in hardware, so unless this is reprogrammable there is no chance to fake negotiation. The next issue is essentially the connection between the Policy Engine/Device Policy Manager and the actual power supply, if that touches something controlled by user SW you have a problem regardless of negotiation.
This whole area is a different question.
The other thing is that devices could lie and cause some problems, that will be caught in compliance testing IF the product goes to compliance testing at all.
Q9 How does the cost and size scale with the power of the device?
A9 When looking at the various parts of the system, most things are meant to “disappear” into an IC where they will take so little space that is is unnoticeable. What remains are the DC separation capacitors, they should be 50V 10nF and can be had in 0402 size at almost no cost and last the zIsolation.
zIsolation has the purpose to separate the big capacitors on the power supply from the signalling on the cable. The default component for this purpose is a 1µH inductor. Up to about 1.3A there are a couple of chip inductors about 1.5mm long at very low cost. At higher currents the saturation point must move up as well and the inductor becomes larger and more expensive.
The deal is that you pay for an inductor that will just satisfy your own current requirement and no more, so this part scales well with current.
Q10 You’re currently working on a USB-PD compliance test equipment. The standard doesn’t specify compliance tests and procedure. Can you detail the qualification procedures for a USB-PD certification, if any?
A10 USB has always made the distinction of specifying the operation and the compliance requirements separately. In due time there will be published a compliance plan which will very likely be an automated sequence of tests.
When the compliance plan is here, the procedure will be that the vendor fills out the checklist describing the UUT you want to put through testing. There will be requirements for providing “whatever” is needed to make the UUT go through all the states it has.
When all of this is in place, the UUT is connected to the tester and the tester guides the person conducting the test to perform the “whatevers” as needed.
You may see one UUT that will offer a certain set of capabilities when on battery power and a different set when powered from the wall, in such a case one “whatever” will be to plug/unplug the wall power.
At the end a report is printed with results.
Let me thank Sten for this interview and for his availability to share his experience and competences.
Owner at S-Carlsen Consulting and Senior Engineer with 25-years+ active experiences in R&D in relation to advanced technology, in many fields and applications.
Having a wide international experience including working in multi-cultural enterprises and being a mentor for foreign and local engineers using net meeting tools as well as face to face meetings. Fast learner of new technologies, can quickly produce results. Using experience from one area of technology to get a new view on a different area and combining different specialities to achieve a better solution.
Never really satisfied with existing solutions, always looking for better ways to design the next product. he is not satisfied with a product that works perfectly technically if the users find it hard to use. Also the full chain needs to be functioning to create a good product, it does not work if only part of the system is working. The customer needs must be satisfied.