I don’t think cost should be the arbiter. With a billion people in need of this, the right solution will get the volume discounts in time. More important IMHO is to find the right solution, taking into account basic raw materials costs, which don’t vary with volume. For example, is it right to require gold for something we’re going to eventually produce billions of?
If USB-C needs the two power pins in order to give it mechanical strength, why stick with the tiny USB-C contacts in the middle?
A different argument applies to the use of the USB-PD2.0 protocols – but even there, I think you will find Africa very distrustful of something so complex and so dependent on the availability of particular chips. To avoid just increasing the dependency of Africa on Silicon Valley and the Chinese, we’d like to move eventually to a situation in which it’s realistic for Africa to start manufacturing locally (and not just using cheap labor for assembling kits). The only way to make this ever in prospect is to keep it simple. The tooling investment for hi-tech is huge (as is buying in the expertise to set it up).
I believe galvanic isolation at the point of delivery is essential for any hi-tech load that has any kind of external port in addition to the power jack (just like it is for LVAC). I’m attaching a short deck on why this is important. (I may have sent this to you before.)
I still believe it’s within the power of an international standards organisation to organise a competition between connector manufacturers to design a connector to meet a functional spec defined by us, and with pre-defined judging criteria. Prototypes could indeed be 3D-printed. The only argument I have heard against this is “We don’t do that sort of thing.”
From: James Gula [mailto:jlgula@papugh.com]
Sent: 15 May 2020 20:45
To: chris.moller@evonet.com.; Martin Jäger
Subject: Re: Wall wart woes - an informative anecdote for ODG
I hear you. But it’s hard to stand in front of a manufacturing juggernaut. I’ve been doing a little more research on the question of if anything can replace the barrel connector AC adapter. As a baseline, if you search for Alibaba for 12V AC adapter, you get prices around $2. On the other hand, the economies of scale for USB-C are really starting to kick in. When last I looked a few months ago, female USB-C sockets were about $1. Now on Alibaba they are less than $0.10. Cables, even 100W, are still around $10 on Amazon retail but on Alibaba they are less than $1. There is even a solution to your fragility concerns with Apple-style mag-safe USB-C (still pretty expensive). A basic 5V 2A USB power adapter now goes for <$1 and even the higher power programmable ones are in the $2 range. Nothing that we do is going to be able to compete with those prices. I think we just need to accept that USB-C is going to power all devices < 100W.
Cypress seems to understand this too. They are offering a USB-C “Barrel Connector Replacement” product. Their device which does the USB-C negotiations goes for $1.23 / 1000. The only other expensive part on their reference design is a FET power switch that goes for around $0.44/1000. These are not complicated devices. Once Chinese manufacturers start creating similar parts, I’m sure the prices will drop substantially.
Having said all that, there are still complications. USB-C is good for charging batteries because if one is willing to wait long enough, even a 5V 2A source will eventually complete the charge. It’s much less clear how devices that need 25W continuously are going to work. People are going to be very confused when they try to plug their 25W device into a 10W adapter and it doesn’t work. The Cypress part has an LED output that shows when there is not enough power available but I’m not aware of any standard for how to indicate this. Maybe part of the answer is that all small devices will come with batteries so they can be used portably though that doesn’t make much sense for a continuous use device like my cable modem. Much wider use of batteries would help on the startup surge problem.
USB-C has another problem relating to ground reference. It works fine as a power cord but won’t work over any distances because of ground shifts. I suggest that we build our architecture model around a 48V floating distribution system using PLC with a USB-C adapter for the device connector (the last mile - well 2m). I still think my old idea about plug and socket that combines USB-C with higher power blade pins is a good idea. A key feature is that it uses the USB connection to prevent arcing on the blade pins - the USB disconnects before the blades and the load must power down before the blades separate. An adapter based on this could even fit inside a cigarette lighter outlet. I still have no clue how to interest the world in this idea - maybe a 3D printed prototype?
The USB-C connector won’t survive 5mins in a domestic electrical context in Africa. Here’s a paper I may have sent you before, on designing a suitable connector.
BTW, I was using my phone as a 3G modem for my laptop, and walking round the college with it to try to find somewhere with good coverage. Unfortunately, the phone slid off the laptop as I was walking. The lead was only a metre long, so the phone didn’t touch the ground – but the jolt on cable was sufficient to break the USB-C connector.
The standard American domestic 115V plug is IMHO about the optimum size for a domestic electrical connector for portable equipment in an uncontrolled environment. Of course, it can be made smaller – but should it?
From: James Gula [mailto:jlgula@papugh.com]
Sent: 13 May 2020 20:41
To: Martin Jäger; chris.moller@evonet.com.
Subject: Wall wart woes - an informative anecdote for ODG
In previous notes, I mentioned that I recently updated my internet connection to high speed which involved the ISP sending someone on site. Part of this involved replacing my cable modem with a shiny new black box. Due to various aches and pain, it’s hard for me to crawl under tables these days so I let the guy crawl around to unplug my old modem, which we set aside and installed the new one. I should have been paying more attention (you can probably see what’s coming).
I’m unusually paranoid about backup so I do 3 kinds of backup on my primary PCs. Apple has a good backup system called Time Machine which I run continuously storing onto a Synology server hidden away in another room (we were once burgled and lost the PCs so I don’t want the backup stolen too). My secondary backup is a file by file backup to an offsite service in case my house burns down. Time Machine is better because it can conveniently do a full system restore including OS etc which I’ve had to do several times do to PC drives failing. Out of extreme paranoia, my 3rd back system is a USB drive which I deliberately leave in a closet most of the time under the theory that if I ever get hit with ransom-ware, clever malware might encrypt my backups but at least I will likely have a snapshot unlikely to be encrypted.
Every couple of weeks Apple nags me that it hasn’t seen my USB drive lately so I brought it in to do a backup. The power cord wasn’t in its usual place so I had to crawl under the table to find it. I found an unconnected wart but I was a little suspicious if it was the right one. I unplugged it and looked at the labels. It didn’t have much, just some safety stuff and couple of other cryptic codes. The drive was clearly labeled that it needs 12V at 1.5A. I measured the voltage on the wart and it was 13.1V. Seemed close enough and the barrel connector fit. What could go wrong?
When I powered up, the drive started beeping. I didn’t even know it could do that. But that was all. Crap. It occurred to me to look at the wart from the cable modem we had set aside. It was clearly labelled with the same brand as the drive and 12V. I was hoping that if the drive was smart enough to beep with bad power it would be smart enough to protect itself. But, of course, when I plugged in the right wart - dead as a dodo. Double crap.
Getting back to ODG. Clearly Chris’s labeling system would have been a big help here. But would I have had the patience to get my phone, do the scans and verify the connectivity? Maybe. Would others? Just requiring these warts to be labeled with the brand and model of the device they are supposed to power would be a big improvement but I don’t see the current US administration doing anything so business unfriendly. Not sure about the UK or EU.
For ODG, I’ve been thinking some about line cords (and plugs :( ) and communications. One straightforward solution would be to standardize on USB-C for appliance connections. I understand the EU has been pushing this with Apple resisting because they make a lot of money licensing their patented Lightning Connector. USB-C wouldn’t guarantee that devices would work because the adapter might not supply enough power but at least nothing would burn out. The problem is price. USB-C has several problems that drive the price too high, basically because it tries to do too much. USB has been evolving in both power and data rate and unfortunately those evolutions have been coupled. Even though USB-C has been out for several years, cable prices are still high because they support the fast data rates. Also the standard trident labeled cable only supports 60W. To get the full 100W potential you have to get a cable made from unobtainium. USB-C chipsets that support power negotiations are still expensive too - probably because they are used primarily in devices that are not as price sensitive compared to things like USB drives and the gazillion other devices using warts.
Could ODG help? Maybe. Once again, the problem is cost. Clearly, appliance makers who use these warts are trying to save every penny. They supply just enough power for the device and they are too cheap to even label the damn things. I think there are several sub issues that ODG would need to address:
• Labeling and branding
• Cables and connectors
• Fault indicators
• Over-voltage protection
• Communications (if any)
Labeling and branding - If we could establish a recognizable, licensed icon or brand indicator, people might gain confidence that if they plugged an ODG source into an ODG load, nothing bad would happen. It still might not work because of insufficient power - see fault indicators. Chris’s bar-codes could be added but supposedly redundant as ODG is just supposed to work.
Cables and Connectors - Inventing some non-standard connector as Chris has suggested would help but that’s daunting challenge. It would have to compete with barrel connectors which cost almost nothing. The wiring too would have to be very inexpensive. One slim possibility. Some time ago I stumbled on something called DIN 45323 (spec attached). It’s basically a slightly odd-shaped barrel connector with a retention feature.The GlobTek part is rated for 5A, 63V. Lumberg makes one but their’s is only rated for 3A, 34V (Chinese versus German ratings?). I don’t know much about connectors but it seems like this could be made for prices similar to regular barrel connectors if we could somehow get the volume high. It’s only two pins so any communications would have to be PLC.
Fault Indicators - Having a consistent way to report a fault would help resolve problems like I had, presuming the device had some fault protection in the first place. If the device had enough protection to survive over voltage, a multi-colored LED could provide a decent information at a small cost increment. I suggest it should at least indicate over voltage and insufficient current with other kinds of device specific faults possible. Safety authorities frown on depending on color alone as an indicator but maybe light could be augmented with patterns etc.
Over-voltage protection - OVP costs but not that much. I’m open to suggestions on how to convince/force appliance manufacturers to include some basic OVP. One of the things that I’ve been thinking about is the interconnection / interaction of 12 and 48V devices. Part of the black-start / source attachment solution will likely involve some kind of toe-in-the-water startup at a voltage that is guaranteed to be safe (< 30V). Speculation: if all ODG devices had the full 48V protection and could communicate, we could do some kind of USB-PD like negotiation even on a bus, starting at 12V and raising the voltage if nobody complains. It seems like the cost increment to make a buck converter work (or at least be safe) across the full 48V range versus just 12V would be modest but then every penny counts.
Communications - Here we are again - how to communicate inexpensively yet robustly? Note that all the issues in this list tend to interact with one another. One observation I would make about USB is that it is over-specified. The tolerances are too tight, required times for voltage shifts are too quick. I think this is part of what makes it expensive. Maybe some kind of looser specifications in communications might help relative to G3-PLC, for example. The tradeoff would be less universality - perhaps inside devices versus outside devices.
Enough for now… kind of rambling for an issue but maybe can be reformulated into an issue.
James Gula
Hi Jim,
I don’t think cost should be the arbiter. With a billion people in need of this, the right solution will get the volume discounts in time. More important IMHO is to find the right solution, taking into account basic raw materials costs, which don’t vary with volume. For example, is it right to require gold for something we’re going to eventually produce billions of?
If USB-C needs the two power pins in order to give it mechanical strength, why stick with the tiny USB-C contacts in the middle?
A different argument applies to the use of the USB-PD2.0 protocols – but even there, I think you will find Africa very distrustful of something so complex and so dependent on the availability of particular chips. To avoid just increasing the dependency of Africa on Silicon Valley and the Chinese, we’d like to move eventually to a situation in which it’s realistic for Africa to start manufacturing locally (and not just using cheap labor for assembling kits). The only way to make this ever in prospect is to keep it simple. The tooling investment for hi-tech is huge (as is buying in the expertise to set it up).
I believe galvanic isolation at the point of delivery is essential for any hi-tech load that has any kind of external port in addition to the power jack (just like it is for LVAC). I’m attaching a short deck on why this is important. (I may have sent this to you before.)
I still believe it’s within the power of an international standards organisation to organise a competition between connector manufacturers to design a connector to meet a functional spec defined by us, and with pre-defined judging criteria. Prototypes could indeed be 3D-printed. The only argument I have heard against this is “We don’t do that sort of thing.”
Best regards,
Chris
===========================================================
From: James Gula [mailto:jlgula@papugh.com] Sent: 15 May 2020 20:45 To: chris.moller@evonet.com.; Martin Jäger Subject: Re: Wall wart woes - an informative anecdote for ODG
I hear you. But it’s hard to stand in front of a manufacturing juggernaut. I’ve been doing a little more research on the question of if anything can replace the barrel connector AC adapter. As a baseline, if you search for Alibaba for 12V AC adapter, you get prices around $2. On the other hand, the economies of scale for USB-C are really starting to kick in. When last I looked a few months ago, female USB-C sockets were about $1. Now on Alibaba they are less than $0.10. Cables, even 100W, are still around $10 on Amazon retail but on Alibaba they are less than $1. There is even a solution to your fragility concerns with Apple-style mag-safe USB-C (still pretty expensive). A basic 5V 2A USB power adapter now goes for <$1 and even the higher power programmable ones are in the $2 range. Nothing that we do is going to be able to compete with those prices. I think we just need to accept that USB-C is going to power all devices < 100W.
Cypress seems to understand this too. They are offering a USB-C “Barrel Connector Replacement” product. Their device which does the USB-C negotiations goes for $1.23 / 1000. The only other expensive part on their reference design is a FET power switch that goes for around $0.44/1000. These are not complicated devices. Once Chinese manufacturers start creating similar parts, I’m sure the prices will drop substantially.
Having said all that, there are still complications. USB-C is good for charging batteries because if one is willing to wait long enough, even a 5V 2A source will eventually complete the charge. It’s much less clear how devices that need 25W continuously are going to work. People are going to be very confused when they try to plug their 25W device into a 10W adapter and it doesn’t work. The Cypress part has an LED output that shows when there is not enough power available but I’m not aware of any standard for how to indicate this. Maybe part of the answer is that all small devices will come with batteries so they can be used portably though that doesn’t make much sense for a continuous use device like my cable modem. Much wider use of batteries would help on the startup surge problem.
USB-C has another problem relating to ground reference. It works fine as a power cord but won’t work over any distances because of ground shifts. I suggest that we build our architecture model around a 48V floating distribution system using PLC with a USB-C adapter for the device connector (the last mile - well 2m). I still think my old idea about plug and socket that combines USB-C with higher power blade pins is a good idea. A key feature is that it uses the USB connection to prevent arcing on the blade pins - the USB disconnects before the blades and the load must power down before the blades separate. An adapter based on this could even fit inside a cigarette lighter outlet. I still have no clue how to interest the world in this idea - maybe a 3D printed prototype?
Jim
===========================================================
On May 14, 2020, at 2:02 AM, chris.moller@evonet.com. chris.moller@evonet.com wrote:
Hi Jim,
The USB-C connector won’t survive 5mins in a domestic electrical context in Africa. Here’s a paper I may have sent you before, on designing a suitable connector.
BTW, I was using my phone as a 3G modem for my laptop, and walking round the college with it to try to find somewhere with good coverage. Unfortunately, the phone slid off the laptop as I was walking. The lead was only a metre long, so the phone didn’t touch the ground – but the jolt on cable was sufficient to break the USB-C connector.
The standard American domestic 115V plug is IMHO about the optimum size for a domestic electrical connector for portable equipment in an uncontrolled environment. Of course, it can be made smaller – but should it?
Best regards,
Chris
===========================================================
From: James Gula [mailto:jlgula@papugh.com] Sent: 13 May 2020 20:41 To: Martin Jäger; chris.moller@evonet.com. Subject: Wall wart woes - an informative anecdote for ODG
In previous notes, I mentioned that I recently updated my internet connection to high speed which involved the ISP sending someone on site. Part of this involved replacing my cable modem with a shiny new black box. Due to various aches and pain, it’s hard for me to crawl under tables these days so I let the guy crawl around to unplug my old modem, which we set aside and installed the new one. I should have been paying more attention (you can probably see what’s coming).
I’m unusually paranoid about backup so I do 3 kinds of backup on my primary PCs. Apple has a good backup system called Time Machine which I run continuously storing onto a Synology server hidden away in another room (we were once burgled and lost the PCs so I don’t want the backup stolen too). My secondary backup is a file by file backup to an offsite service in case my house burns down. Time Machine is better because it can conveniently do a full system restore including OS etc which I’ve had to do several times do to PC drives failing. Out of extreme paranoia, my 3rd back system is a USB drive which I deliberately leave in a closet most of the time under the theory that if I ever get hit with ransom-ware, clever malware might encrypt my backups but at least I will likely have a snapshot unlikely to be encrypted.
Every couple of weeks Apple nags me that it hasn’t seen my USB drive lately so I brought it in to do a backup. The power cord wasn’t in its usual place so I had to crawl under the table to find it. I found an unconnected wart but I was a little suspicious if it was the right one. I unplugged it and looked at the labels. It didn’t have much, just some safety stuff and couple of other cryptic codes. The drive was clearly labeled that it needs 12V at 1.5A. I measured the voltage on the wart and it was 13.1V. Seemed close enough and the barrel connector fit. What could go wrong?
When I powered up, the drive started beeping. I didn’t even know it could do that. But that was all. Crap. It occurred to me to look at the wart from the cable modem we had set aside. It was clearly labelled with the same brand as the drive and 12V. I was hoping that if the drive was smart enough to beep with bad power it would be smart enough to protect itself. But, of course, when I plugged in the right wart - dead as a dodo. Double crap.
Getting back to ODG. Clearly Chris’s labeling system would have been a big help here. But would I have had the patience to get my phone, do the scans and verify the connectivity? Maybe. Would others? Just requiring these warts to be labeled with the brand and model of the device they are supposed to power would be a big improvement but I don’t see the current US administration doing anything so business unfriendly. Not sure about the UK or EU.
For ODG, I’ve been thinking some about line cords (and plugs :( ) and communications. One straightforward solution would be to standardize on USB-C for appliance connections. I understand the EU has been pushing this with Apple resisting because they make a lot of money licensing their patented Lightning Connector. USB-C wouldn’t guarantee that devices would work because the adapter might not supply enough power but at least nothing would burn out. The problem is price. USB-C has several problems that drive the price too high, basically because it tries to do too much. USB has been evolving in both power and data rate and unfortunately those evolutions have been coupled. Even though USB-C has been out for several years, cable prices are still high because they support the fast data rates. Also the standard trident labeled cable only supports 60W. To get the full 100W potential you have to get a cable made from unobtainium. USB-C chipsets that support power negotiations are still expensive too - probably because they are used primarily in devices that are not as price sensitive compared to things like USB drives and the gazillion other devices using warts.
Could ODG help? Maybe. Once again, the problem is cost. Clearly, appliance makers who use these warts are trying to save every penny. They supply just enough power for the device and they are too cheap to even label the damn things. I think there are several sub issues that ODG would need to address:
• Labeling and branding • Cables and connectors • Fault indicators • Over-voltage protection • Communications (if any)
Labeling and branding - If we could establish a recognizable, licensed icon or brand indicator, people might gain confidence that if they plugged an ODG source into an ODG load, nothing bad would happen. It still might not work because of insufficient power - see fault indicators. Chris’s bar-codes could be added but supposedly redundant as ODG is just supposed to work.
Cables and Connectors - Inventing some non-standard connector as Chris has suggested would help but that’s daunting challenge. It would have to compete with barrel connectors which cost almost nothing. The wiring too would have to be very inexpensive. One slim possibility. Some time ago I stumbled on something called DIN 45323 (spec attached). It’s basically a slightly odd-shaped barrel connector with a retention feature.The GlobTek part is rated for 5A, 63V. Lumberg makes one but their’s is only rated for 3A, 34V (Chinese versus German ratings?). I don’t know much about connectors but it seems like this could be made for prices similar to regular barrel connectors if we could somehow get the volume high. It’s only two pins so any communications would have to be PLC.
Fault Indicators - Having a consistent way to report a fault would help resolve problems like I had, presuming the device had some fault protection in the first place. If the device had enough protection to survive over voltage, a multi-colored LED could provide a decent information at a small cost increment. I suggest it should at least indicate over voltage and insufficient current with other kinds of device specific faults possible. Safety authorities frown on depending on color alone as an indicator but maybe light could be augmented with patterns etc.
Over-voltage protection - OVP costs but not that much. I’m open to suggestions on how to convince/force appliance manufacturers to include some basic OVP. One of the things that I’ve been thinking about is the interconnection / interaction of 12 and 48V devices. Part of the black-start / source attachment solution will likely involve some kind of toe-in-the-water startup at a voltage that is guaranteed to be safe (< 30V). Speculation: if all ODG devices had the full 48V protection and could communicate, we could do some kind of USB-PD like negotiation even on a bus, starting at 12V and raising the voltage if nobody complains. It seems like the cost increment to make a buck converter work (or at least be safe) across the full 48V range versus just 12V would be modest but then every penny counts.
Communications - Here we are again - how to communicate inexpensively yet robustly? Note that all the issues in this list tend to interact with one another. One observation I would make about USB is that it is over-specified. The tolerances are too tight, required times for voltage shifts are too quick. I think this is part of what makes it expensive. Maybe some kind of looser specifications in communications might help relative to G3-PLC, for example. The tradeoff would be less universality - perhaps inside devices versus outside devices.
Enough for now… kind of rambling for an issue but maybe can be reformulated into an issue. James Gula
Galvanic Isolation.pptx Domestic LVDC Connector Use Case.docx