Any way to safely test if an AC-DC converter can accept DC input?

[FyKnight]

Hi, I'm new, please correct me gently if I do something wrong. I did look to see if this question has been asked before but didn't find anything. Also I'm thrilled to find this forum!

I have a 7 kW AC-DC converter that I would like to feed DC directly to instead of AC. Yes, it's the OnBoard Charger in my EV. I'll need to charge it at an off-grid location, where I'll build a dedicated system just for this purpose. It seems silly to have a big inverter there if it can be avoided.

It could well be possible, right? If I keep to half the max current, then even though only one pair of diodes in the rectifier will be used, they shouldn't waste more power than under AC at max current. By the time we get to a capacitor it'll be no different to AC. The big question mark for me though is what will the PFC boost converter algorithm do?

So before I go any further with this thinking I'd like to test it. Is there any way I can do so without a large risk of damage to the OBC? All I can think of is to use the minimum voltage the OBC accepts (85 Vrms AC, so I guess 120 V DC?), at the lowest current (6 A, i.e. getting the EVSE to tell it to only draw 6 A) and try it for just a few seconds.

I'd also be really keen to hear if someone has done this before or even used a modified sine wave inverter to charge their EV (since that is nothing like a sine wave at all)

Thanks!

[mikeselectricstuff]

Hard to see how it would damage it, shouldn't be too hard to use a low current DC supply to test off-load.
Might be useful to take a look at the circuitry to see if there any obvious potential issues.
 

[NiHaoMike]

Perhaps just use the DC charge mode if there is one? It's often called "fast charging" but it most likely will also work at somewhat slower rates. https://openinverter.org/forum/viewforum.php?f=17

[FyKnight]

Thanks very much for your replies!

Hard to see how it would damage it, shouldn't be too hard to use a low current DC supply to test off-load.
Might be useful to take a look at the circuitry to see if there any obvious potential issues.

OK interesting thank you! Will using a low current DC supply get much data? The OBC minimum configurable draw is 6 A, so it would just overload a lesser supply very quickly wouldn't it?

For the circuitry, would a photo on its own be any use or would I need to go and reverse engineer the actual circuit? I can probably manage to do that but if it isn't a massive risk perhaps I'll just try it and see.

Perhaps just use the DC charge mode if there is one? It's often called "fast charging" but it most likely will also work at somewhat slower rates. https://openinverter.org/forum/viewforum.php?f=17

There is indeed a CCS fast charge mode in my EV, and that was my initial plan. I couldn't find any EVSEs under AU$50K that support DC charging on the CCS pins — they are all designed for fast charging. So I have to build one myself.

The first problem with this mode is that unlike AC charging which is controlled by a simple 1 kHz PWM signal, the DC charging uses a nightmare combination starting with powerline communications protocols over the same PWM control line and getting worse from there: (ROBO) OFDM, Ethernet, "Sounding packets", IP, TCP, TLS, HTTP and the cherry on top: XML! It's disgusting, and you need a beefy microcontroller to run it, but there are modules available and I am trying to order one (fingers crossed *).

The bigger hurdle with this approach though is that the battery is directly connected to the DC pins in the socket, and it's up to the EVSE to generate the voltage required to charge it, so that requires a DC-DC converter of similar complexity to the inverter that I'm trying to eliminate. It's still more efficient overall so worth doing, and I have a couple of these on order too.

So yeah, if I can reuse the hardware that I'm already carrying around, that would be cheapest, easiest and likely most efficient solution. I wish I'd realised this before spending so long going down the HVDC and CCS rabbit holes. Also, thanks very much for the openinverter link! Heaps of great info on that forum!

*: Probably easier than doing CCS is to just put a T junction on the HV lines from the battery and drive the contactors directly... if only the connectors didn't cost $200+

[mikeselectricstuff]

Perhaps just use the DC charge mode if there is one? It's often called "fast charging" but it most likely will also work at somewhat slower rates. https://openinverter.org/forum/viewforum.php?f=17

The problem with that is that you need to do the current regulation externally - the CCS port is just a pair of contactors away from  the battery terminals

[mikeselectricstuff]

Thanks very much for your replies!

Hard to see how it would damage it, shouldn't be too hard to use a low current DC supply to test off-load.
Might be useful to take a look at the circuitry to see if there any obvious potential issues.

OK interesting thank you! Will using a low current DC supply get much data? The OBC minimum configurable draw is 6 A, so it would just overload a lesser supply very quickly wouldn't it?

The 6A is just what the EVSE advertises, you could use a dummy load to draw less for testing

For the circuitry, would a photo on its own be any use or would I need to go and reverse engineer the actual circuit? I can probably manage to do that but if it isn't a massive risk perhaps I'll just try it and see.

You may or may not see anything conclusive, but if there is, for example an AC current transformer, or (fairly unlikely) a conventional transformer to generate an aux supply that would indicate possible issues.

[FyKnight]

OK interesting thank you! Will using a low current DC supply get much data? The OBC minimum configurable draw is 6 A, so it would just overload a lesser supply very quickly wouldn't it?

The 6A is just what the EVSE advertises, you could use a dummy load to draw less for testing

As I understand it, it's what the EVSE advertises, but it's the OBC that decides how much to pull. And it's the OBC that I'm trying to test. (According to the spec it has 5 sec to adjust its current to what the pilot signal says after a change). I mean the load (the battery) can draw hundreds (thousands?) of amps if it wanted. There's probably something I don't understand here, so please explain it to me if I'm wrong! Ah wait, should I be disconnecting it and replacing the battery with something of higher resistance? I'd be worried that would be bad for the OBC.

Hang on though, this has given me an idea, which might be what you meant all along. If I charge the battery to almost full then the OBC internally ramps down and draws less and less. I don't know if it ever gets under 6 A but maybe it could! Yeah that'd be a good and safer way to test it definitely! Sweet. And bonus, if I break it, at least I'll have a full charge to drive somewhere for a fix 

For the circuitry, would a photo on its own be any use or would I need to go and reverse engineer the actual circuit? I can probably manage to do that but if it isn't a massive risk perhaps I'll just try it and see.

You may or may not see anything conclusive, but if there is, for example an AC current transformer, or (fairly unlikely) a conventional transformer to generate an aux supply that would indicate possible issues.

Ohh yeah good point! Given the size and power of the OBC I can pretty much rule out a main path transformer, but an auxiliary might be possible hmmm. I guess I'd better open it up and have a squizz! Thanks again.

[richard.cs]

An AC current transformer seems likely to me, the OBC has to have some way to measure input current and this would be a normal way to do it. Even if it used Hall sensors for current measurement the software might be configured to reject any DC component. If it can't measure input current it may throw an error and shutdown, or keep trying to draw more power (though you should be able to prevent damage easily enough).

Personally I would probably look at either the lowest-cost way to supply it with "good enough" AC (e.g. square wave), or move the the CCS port.

Some cars also do other tests on the AC port, e.g. a Renault Zoe measures the impedance of the line-earth loop and refuses to charge if it is over 100R.

For the CCS port you'd need at least enough communication to persuade the car to close its contactor, but perhaps as you know the model of the car and its battery type you don't actually need the car to tell you what it needs like a general purpose charging point would do? I am thinking that once the contactor is closed you could get away with a current limited voltage source of manually configured values, and that it would work to arbitrarily low charging power.

[mikeselectricstuff]

For the CCS port you'd need at least enough communication to persuade the car to close its contactor, but perhaps as you know the model of the car and its battery type you don't actually need the car to tell you what it needs like a general purpose charging point would do? I am thinking that once the contactor is closed you could get away with a current limited voltage source of manually configured values, and that it would work to arbitrarily low charging power.

The problem with CCS is the car will think it's connected to a charger, and will probably expect the current supplied to match what it has asked for. Hard to know how it will behave in practice, and likely to vary between different cars.
 

[mikeselectricstuff]

OK interesting thank you! Will using a low current DC supply get much data? The OBC minimum configurable draw is 6 A, so it would just overload a lesser supply very quickly wouldn't it?

The 6A is just what the EVSE advertises, you could use a dummy load to draw less for testing

As I understand it, it's what the EVSE advertises, but it's the OBC that decides how much to pull. And it's the OBC that I'm trying to test. (According to the spec it has 5 sec to adjust its current to what the pilot signal says after a change). I mean the load (the battery) can draw hundreds (thousands?) of amps if it wanted. There's probably something I don't understand here, so please explain it to me if I'm wrong! Ah wait, should I be disconnecting it and replacing the battery with something of higher resistance? I'd be worried that would be bad for the OBC.

Sorry I was getting confused between the OBC and the EVSE!

Hang on though, this has given me an idea, which might be what you meant all along. If I charge the battery to almost full then the OBC internally ramps down and draws less and less. I don't know if it ever gets under 6 A but maybe it could! Yeah that'd be a good and safer way to test it definitely! Sweet. And bonus, if I break it, at least I'll have a full charge to drive somewhere for a fix 

yes, that would be a way to limit the output current. Some cars also have a user-settable way to limit AC charge current

[richard.cs]

For the CCS port you'd need at least enough communication to persuade the car to close its contactor, but perhaps as you know the model of the car and its battery type you don't actually need the car to tell you what it needs like a general purpose charging point would do? I am thinking that once the contactor is closed you could get away with a current limited voltage source of manually configured values, and that it would work to arbitrarily low charging power.

The problem with CCS is the car will think it's connected to a charger, and will probably expect the current supplied to match what it has asked for. Hard to know how it will behave in practice, and likely to vary between different cars.

I was thinking about this, but there has to be a method by which the car gets less than it has asked for and is happy, it is common for chargers to be rated at less than the car, or to wind down if charging several cars. The question is if the charger has to tell the car it's giving it less, or if the request from the car is just treated as a maximum.

Plan C of course is a direct battery connection not via CCS. 

[richard.cs]

Personally I would probably look at either the lowest-cost way to supply it with "good enough" AC (e.g. square wave)...

Thinking about this a bit more, what DC source do you actually have available? There might be something straightforward you can do.

Also, note that whilst the lowest charging current in the standard is 6A, cars are not required to support it. e.g. the Zoe minimum current is 10A (was 16 A on early models prior to a software update).

[sandalcandal]

I have some experience I can share on EV charging systems.

As others have mentioned, there is a good chance your EV's OBC has safety and monitoring systems which assume AC input to the AC charging ports such as an AC current transformer and potentially an auxiliary transformer. Best, non-working case: the vehicle just refuses charge, possibly enters a locked-up fail state. Worst case: there is some auxiliary transformer which burns up and destroys your OBC when you try feeding in DC. I don't think there is a safe way to feed in a DC and validate function without running into issues. OBC systems have contactors which should disconnect all power circuitry until AC charging protocols have been completed. Once pre-charge communication and checks are done, the EV will close contactors to the OBC then it will (usually) do a "soft start" but the speed and power of this "soft start" would not give you a safe window to determine if there is a low auxiliary transformer burning up or not.

The only "safe" method I can think of would be opening up and reverse engineering the OBC. Can you share what model EV you have? or are you trying to make a general charging system?

As for suggestions on DC fast charging, as Mike mentioned, you need a current limiting DC-DC converter. DC fast charging connects the EV's battery directly to the fast charging port. Current is up to the EVSE to control. All the EV can do is try open its internal contactors or blow a fuse if current control is incorrect/not present. Furthermore, AC charging protocols are completely different to CCS (and CHAdeMO) DC charging protocols which are non-trivial to implement (you thought USB PD was a pain? HA!). CCS in particular requires a homeplug green PHY modem IC for communication which is not something you can just get. IC are all behind NDAs and dev agreements AFAIK. There are CCS (and some CHAdeMO) EVSE communication modules available off-the-shelf which mean you don't have to direct dev work on the communication subsystem but the dev kits aren't particularly cheap, also behind NDAs and they aren't "just works" out of the box either (and unlikely they'd sell to hobbyists too). They need to be integrated with control systems for the DC-DC converter and other safety and monitoring systems.

Again with any DC charging, you'll need a current regulating DC-DC converter of sufficient power capability. You're unlikely to find anything off the shelf for cheaper than an equivalent power inverter. You then also get the benefits over being able to run other mains appliances off the inverter.

My suggestion: just use an inverter, you can probably get away with a "modified sine" type.

Edit: Also definitely would not recommend bypassing safety functions and trying to connect directly to the EV battery obviously.
Edit2: If the OBC does have a DC incompatible device, trying to replace/eliminate that would be another option that would be much cheaper as well as significantly easier than implementing a DC charging system (that doesn't blow up and is safe).

[sandalcandal]

Hang on though, this has given me an idea, which might be what you meant all along. If I charge the battery to almost full then the OBC internally ramps down and draws less and less. I don't know if it ever gets under 6 A but maybe it could! Yeah that'd be a good and safer way to test it definitely! Sweet. And bonus, if I break it, at least I'll have a full charge to drive somewhere for a fix 

yes, that would be a way to limit the output current. Some cars also have a user-settable way to limit AC charge current

From what I've seen, the EV disconnects and stops charging (at something like a few hundred watts) before power reaches something low enough that I'd feel safe not burning up an auxiliary transformer (5W or less). No harm in trying and seeing what your EV does though.

[mikeselectricstuff]

All the EV can do is try open its internal contactors or blow a fuse if current control is incorrect/not present.

..that would be the ~600 amp pack fuse..!

[sandalcandal]

All the EV can do is try open its internal contactors or blow a fuse if current control is incorrect/not present.

..that would be the ~600 amp pack fuse..!

Hence why I warn against it! A DC-DC current regulating converter with properly implemented safety systems is a MUST!

[FyKnight]

An AC current transformer seems likely to me, the OBC has to have some way to measure input current and this would be a normal way to do it. Even if it used Hall sensors for current measurement the software might be configured to reject any DC component. If it can't measure input current it may throw an error and shutdown, or keep trying to draw more power (though you should be able to prevent damage easily enough).

It could just as well measure current later in the path, which it needs to do anyway to not charge the battery faster than it wants to. I've noticed that the car's measurement is slightly below what the EVSE says too.

Personally I would probably look at either the lowest-cost way to supply it with "good enough" AC (e.g. square wave), or move the the CCS port.

Square wave is something I've considered. It'd just require a few FETs (and their drivers probably) plus a simple microcontroller (or even the trusty old 555?). That'll be safe enough for any transformers / coils then, but I don't know what the PFC algo would do with such a waveform.

Some cars also do other tests on the AC port, e.g. a Renault Zoe measures the impedance of the line-earth loop and refuses to charge if it is over 100R.

Interesting, I don't think my car has gone overboard with the tests though, it seems to be pretty permissive compared to others...

For the CCS port you'd need at least enough communication to persuade the car to close its contactor, but perhaps as you know the model of the car and its battery type you don't actually need the car to tell you what it needs like a general purpose charging point would do? I am thinking that once the contactor is closed you could get away with a current limited voltage source of manually configured values, and that it would work to arbitrarily low charging power.

The needs of the car vary over time (depending on SoC) and since I'm going to all this effort, I may as well let others in the same situation benefit like me. That said my charging power will definitely be under 10 kW, way lower than any fast DC charger.

[sandalcandal]

If you're really still interested in CCS, you can get some documentation/information packs from CharIN (the association in charge of CCS) on request for free https://www.charin.global/ The info pack has pretty complete information on the CCS protocol and processes includes some detailed slides on the charging process as well as safe guards. Not quite 100% of what you need for a from scratch implementation due to needing some cited standards but pretty good none-the-less.

Edit: If you look for "Design_Guide_Combined_Charging_System" there's an older version floating around on the internet but the contents looks the same as the version I have. Both are dated 2015-06-02.

[richard.cs]

Edit: Also definitely would not recommend bypassing safety functions and trying to connect directly to the EV battery obviously.

Well, creating a battery connection with equivalent safety to the CSS connector but without the godawful communications protocol doesn't feel like a hugely difficult task. Yes, it requires some technical knowledge, but no more so than anything else involving DC charging. One could certainly imagine adding a moderate-current-rated connector* to the battery circuit with interlocking, fusing, etc. The remaining question being if charging without the car knowing about it is likely to upset it or the BMS.

*Something from the IEC 60309 family with a pilot contact perhaps? Or a type 2 in the uncommonly used but standardised DC mode?

[sandalcandal]

Edit: Also definitely would not recommend bypassing safety functions and trying to connect directly to the EV battery obviously.

Well, creating a battery connection with equivalent safety to the CSS connector but without the godawful communications protocol doesn't feel like a hugely difficult task. Yes, it requires some technical knowledge, but no more so than anything else involving DC charging. One could certainly imagine adding a moderate-current-rated connector* to the battery circuit with interlocking, fusing, etc. The remaining question being if charging without the car knowing about it is likely to upset it or the BMS.

*Something from the IEC 60309 family with a pilot contact perhaps? Or a type 2 in the uncommonly used but standardised DC mode?

Certainly not impossible but EV batteries aren't something to be taken lightly. "If you need to ask then you shouldn't be trying" situation. More so when the impetus seems to be avoiding buying a relatively cheap inverter which would be much safer, simpler and almost guaranteed to work.

Edit: Pre-charge, pre-charge insulation checking, insulation monitoring, connector temperature monitoring and contactor welding checks would be a non-exhaustive list of safety features not mentioned. The standard IEC 61851.23:2021 Annex CC describes CCS as a 18 step process for normal startup to shut down, that's with all the communication phases lumped into single steps.
Edit2: Communication integrity checking in another one. Done in CCS as the Signal Level Attenuation Characterization (SLAC) protocol.

[FyKnight]

The problem with CCS is the car will think it's connected to a charger, and will probably expect the current supplied to match what it has asked for. Hard to know how it will behave in practice, and likely to vary between different cars.

I was thinking about this, but there has to be a method by which the car gets less than it has asked for and is happy, it is common for chargers to be rated at less than the car, or to wind down if charging several cars. The question is if the charger has to tell the car it's giving it less, or if the request from the car is just treated as a maximum.

Yes, hopefully it's a maximum, when I use a 50 kW fast charger my car asks for 86 kW but it doesn't get it. There is a minimum field though. I don't know what my car will do until I try it. Of course this is the same port used for V2G, i.e. negative charging current so I hope it will be okay. Here is an extract from the parsed relevant EXI request (I believe) in OpenV2G:

Code: [Select]

struct iso2DC_EVChargeParameterType {
    uint32_t DepartureTime ;
    struct iso2PhysicalValueType EVMaximumChargePower ;
    struct iso2PhysicalValueType EVMinimumChargePower ;
    struct iso2PhysicalValueType EVMaximumChargeCurrent ;
    struct iso2PhysicalValueType EVMinimumChargeCurrent ;
    struct iso2PhysicalValueType EVMaximumVoltage ;
    struct iso2PhysicalValueType EVTargetEnergyRequest ;
    struct iso2PhysicalValueType EVMaximumEnergyRequest ;
    struct iso2PhysicalValueType EVMinimumEnergyRequest ;
    int8_t CurrentSOC ;
    int8_t TargetSOC ;
    int8_t BulkSOC ;
};


Plan C of course is a direct battery connection not via CCS. 

That was Plan A! But the connectors are soooo expensive, and I would still need to drive the contactors myself. So I may as well do it nicely. Also it'll be useful to more people if it uses an existing port.

[FyKnight]

Personally I would probably look at either the lowest-cost way to supply it with "good enough" AC (e.g. square wave)...

Thinking about this a bit more, what DC source do you actually have available? There might be something straightforward you can do.

Also, note that whilst the lowest charging current in the standard is 6A, cars are not required to support it. e.g. the Zoe minimum current is 10A (was 16 A on early models prior to a software update).

My car is an MG ZS EV and it does support 6A, I've already tested it. The stock granny charger that came with it is only 8A too!

The DC sources are solar or battery / supercap bank. The battery will be much smaller than the car traction battery of course.

So the big underlying idea if it can take DC or square wave is to connect the solar (of sufficient voltage and current, which effectively means ~1 kW) directly to the AC port (through an EVSE naturally), and use the car's OBC as the coarse MPPT by varying the current. Yes, if a cloud goes over and it drops out then the charging will stop. Buffering with capacitors or battery would be expensive and ruin the MPPT idea. OK you could have a battery on the side that is switched in but now we have a complicated system again.

[tunk]

Maybe you could contact the maker of your car?
The reason for electrical cars are that they are environmentally friendly.
Charging them off-grid with solar, wind etc. is even more friendly, so it
should be in their interest to help you with this.

[FyKnight]

I have some experience I can share on EV charging systems.

As others have mentioned, there is a good chance your EV's OBC has safety and monitoring systems which assume AC input to the AC charging ports such as an AC current transformer and potentially an auxiliary transformer. Best, non-working case: the vehicle just refuses charge, possibly enters a locked-up fail state. Worst case: there is some auxiliary transformer which burns up and destroys your OBC when you try feeding in DC. I don't think there is a safe way to feed in a DC and validate function without running into issues. OBC systems have contactors which should disconnect all power circuitry until AC charging protocols have been completed. Once pre-charge communication and checks are done, the EV will close contactors to the OBC then it will (usually) do a "soft start" but the speed and power of this "soft start" would not give you a safe window to determine if there is a low auxiliary transformer burning up or not.

Awesome, thank you very much for your insight! Yeah that matches my understanding of the process and is exactly what I'm afraid of. Do you think it would be safe enough to test with a square wave input? I have even thought of having the lid to the OBC open and watching it through an IR camera

The only "safe" method I can think of would be opening up and reverse engineering the OBC. Can you share what model EV you have? or are you trying to make a general charging system?

I would like it to be more general but if it is only for my model car then oh well. It's the cheapest EV in Australia, I'm sure you know the one: the MG ZS EV.

As for suggestions on DC fast charging, as Mike mentioned, you need a current limiting DC-DC converter. DC fast charging connects the EV's battery directly to the fast charging port. Current is up to the EVSE to control. All the EV can do is try open its internal contactors or blow a fuse if current control is incorrect/not present. Furthermore, AC charging protocols are completely different to CCS (and CHAdeMO) DC charging protocols which are non-trivial to implement (you thought USB PD was a pain? HA!). CCS in particular requires a homeplug green PHY modem IC for communication which is not something you can just get. IC are all behind NDAs and dev agreements AFAIK. There are CCS (and some CHAdeMO) EVSE communication modules available off-the-shelf which mean you don't have to direct dev work on the communication subsystem but the dev kits aren't particularly cheap, also behind NDAs and they aren't "just works" out of the box either (and unlikely they'd sell to hobbyists too). They need to be integrated with control systems for the DC-DC converter and other safety and monitoring systems.

Thank you for the overview. I've been researching it for months too, though it sounds like you're more familiar with the systems than that. I have two models of such DC-DC converters on the way, and hopefully a HomePlug GP EVSE module too — it's just a matter of buying in sufficient volume. I want to make this happen, and if I have to build out a company and employ EEs then I will ... but not until necessary (I'm a computer engineer, I've studied electrical engineering but have little experience on the power side. I have several failed startups under my belt too haha). However, if I don't need to go down the CCS path then you can bet I'm going to avoid it!!

Again with any DC charging, you'll need a current regulating DC-DC converter of sufficient power capability. You're unlikely to find anything off the shelf for cheaper than an equivalent power inverter. You then also get the benefits over being able to run other mains appliances off the inverter.

I appreciate the sentiment and would have said the same in general. However this is primarily about efficiency, exploring the options and hopefully making a model system, not so much up-front cost. I already have an inverter (it's huge) and battery system that I can charge with. I want to do it better!

My suggestion: just use an inverter, you can probably get away with a "modified sine" type.

That's excellent. Modified sine is usually just square wave with a gap at zero between the cycles, as I understand it. If that can work then the square wave might work and we are good to go!

Edit: Also definitely would not recommend bypassing safety functions and trying to connect directly to the EV battery obviously.
Edit2: If the OBC does have a DC incompatible device, trying to replace/eliminate that would be another option that would be much cheaper as well as significantly easier than implementing a DC charging system (that doesn't blow up and is safe).

I don't want to get anywhere near the 450V believe me. Ohhh interesting, yeah! Hmmm... though that might work for me, it doesn't have the benefit for many other people...

[sandalcandal]

I want to make this happen, and if I have to build out a company and employ EEs then I will ... but not until necessary (I'm a computer engineer, I've studied electrical engineering but have little experience on the power side. I have several failed startups under my belt too haha). However, if I don't need to go down the CCS path then you can bet I'm going to avoid it!!

Maybe you should come join us then we're based in Sydney/Canberra.

[FyKnight]

If you're really still interested in CCS, you can get some documentation/information packs from CharIN (the association in charge of CCS) on request for free https://www.charin.global/ The info pack has pretty complete information on the CCS protocol and processes includes some detailed slides on the charging process as well as safe guards. Not quite 100% of what you need for a from scratch implementation due to needing some cited standards but pretty good none-the-less.

Edit: If you look for "Design_Guide_Combined_Charging_System" there's an older version floating around on the internet but the contents looks the same as the version I have. Both are dated 2015-06-02.

Thanks again sandalcandal, if it's the one with the Euro car manufacturer logos at the top of every page then yes, I found that document a couple of months ago when I started this journey, it's a really good overview. Is there more detail in the info pack? i.e. actual standards? Funnily enough I was on their site again today downloading a few things I'd missed.

I'm not against buying the standards either (especially in Estonian Euros ) but if this direct DC or square wave idea works out then I won't have to.

I really appreciate your replies as well as all the other responders to this thread. What a great community!

[mikeselectricstuff]

I just had a look at the board from an old OBC I have here - I think it's from a BMW PHEV.
I can't see anything that would preclude it running on DC. It doesn't have a current transformer, only a sense resistor after the input rectifier. This makes sense as it can derive AC input current from this, and it's smaller and cheaper than a CT.

I suspect the most likely obstacle is a software-controlled PFC wanting to see AC, though I can't see why it would care, unless it's using something like synchronous rectification for efficiency. Should be fairly obvious from the board if this is the case.

[sandalcandal]

Some notes on compatibility with square wave inverters with respect to standards. J1772 clause 4.6.1.4 stipulates harmonic distortion immunity specified in UL 2231-2 (can be view free here: https://standardscatalog.ul.com/ProductDetail.aspx?productId=UL2231-2) J2894-1 "Power Quality Requirements for Plug-In Electric Vehicle Chargers" also recommends THD of 10% [max] but practically speaking, apart from possible but unlikely noise [and PFC control loop] issues I don't see any reason they shouldn't work.

However, if efficiency is the concern, current state of the art gets over 99% efficiency with "pure sine" output using WBG (wide band gap) semiconductors and a totem pole topology e.g. https://training.ti.com/demo-900-v-5kw-bidirectional-ac-dc-converter-gan Obviously there's a bit of a cost premium but if you're making something new, might as well.

[sandalcandal]

Thanks again sandalcandal, if it's the one with the Euro car manufacturer logos at the top of every page then yes, I found that document a couple of months ago when I started this journey, it's a really good overview. Is there more detail in the info pack? i.e. actual standards? Funnily enough I was on their site again today downloading a few things I'd missed.

I'm not against buying the standards either (especially in Estonian Euros ) but if this direct DC or square wave idea works out then I won't have to.

I really appreciate your replies as well as all the other responders to this thread. What a great community!

Yep, ~130 page long slide deck with some EU car groups at the top of every page. That's really the main one with all the juicy info, the other documents are mostly formalising documents citing standards and branding guidelines. The fine engineering details e.g. numerical values, timing and other parameters are all in the pay walled standards unfortunately.

Good you know about the Estonians (I hope it doesn't get shut down).

[NiHaoMike]

Again with any DC charging, you'll need a current regulating DC-DC converter of sufficient power capability. You're unlikely to find anything off the shelf for cheaper than an equivalent power inverter. You then also get the benefits over being able to run other mains appliances off the inverter.

Easily available for about $100 in the form of the buck/boost part of a Prius inverter, good for 10-20kW continuous. You'll also get two 3 phase inverter outputs and a 12V DC/DC converter.

[sandalcandal]

Again with any DC charging, you'll need a current regulating DC-DC converter of sufficient power capability. You're unlikely to find anything off the shelf for cheaper than an equivalent power inverter. You then also get the benefits over being able to run other mains appliances off the inverter.

Easily available for about $100 in the form of the buck/boost part of a Prius inverter, good for 10-20kW continuous. You'll also get two 3 phase inverter outputs and a 12V DC/DC converter.

That sounds like a nice idea. I didn't think about using a salvaged inverter. Are there any good resources on getting a Prius inverter working outside the car?

[NiHaoMike]

https://openinverter.org/forum/viewforum.php?f=14
Note that although the low side cap limits the working voltage to 270V or so, you can get around it by connecting the input between the high side and low side. You'll also have to change out the board with the control ASICs for one that uses more generic control logic.

[FyKnight]

I just had a look at the board from an old OBC I have here - I think it's from a BMW PHEV.
I can't see anything that would preclude it running on DC. It doesn't have a current transformer, only a sense resistor after the input rectifier. This makes sense as it can derive AC input current from this, and it's smaller and cheaper than a CT.

Oh awesome thanks very much! I'll try to open mine up and have a look this arvo.

I suspect the most likely obstacle is a software-controlled PFC wanting to see AC, though I can't see why it would care, unless it's using something like synchronous rectification for efficiency. Should be fairly obvious from the board if this is the case.

So if I see a regular diode FULL BRIDGE rectifier (oh wrong channel, sorry!) then it should be right?

[FyKnight]

Some notes on compatibility with square wave inverters with respect to standards. J1772 clause 4.6.1.4 stipulates harmonic distortion immunity specified in UL 2231-2 (can be view free here: https://standardscatalog.ul.com/ProductDetail.aspx?productId=UL2231-2) J2894-1 "Power Quality Requirements for Plug-In Electric Vehicle Chargers" also recommends THD of 10% [max] but practically speaking, apart from possible but unlikely noise [and PFC control loop] issues I don't see any reason they shouldn't work.

OK brilliant! Since it's a stand-alone system I'm not worried about regular noise affecting other devices, although I'm a little concerned about spikes when switching.

However, if efficiency is the concern, current state of the art gets over 99% efficiency with "pure sine" output using WBG (wide band gap) semiconductors and a totem pole topology e.g. https://training.ti.com/demo-900-v-5kw-bidirectional-ac-dc-converter-gan Obviously there's a bit of a cost premium but if you're making something new, might as well.

The thing is, apart from the cost and weight, I have to factor in the OBC inefficiency. And the battery efficiency (intra cycle hopefully), then the MPPT efficiency. It doesn't just add up, it multiplies out to a pretty bad deal.

Yes I looked into using a batteryless inverter but they seem to support well less than their regular rated load (and cut out) when no battery is present, and initial capacitor charging current knocks them over too. It also isn't clear if they do MPPT in batteryless mode, and finally I don't know how to determine the used vs available power in order to ramp the current limit (request) up and down. So I discarded that option to focus on direct DC, and in the meantime went with a regular inverter + a couple of MPPT charge controllers and a high C battery to get something working (like everyone else). This was also informed by doing a couple of measurements of my OBC efficiency which came out to a total loss of 300 - 400 W (just comparing SoC after a time vs input energy). I can't see how that can true though after learning more about OBC design so I'll revisit it. Maybe the 12 V DC-DC charger was running at that time...

[FyKnight]

https://openinverter.org/forum/viewforum.php?f=14
Note that although the low side cap limits the working voltage to 270V or so, you can get around it by connecting the input between the high side and low side. You'll also have to change out the board with the control ASICs for one that uses more generic control logic.

That is a pretty cool option hardware-wise, but implementing the control logic could be a bit hit and miss and take a good while to get right. I'd probably be better off spending the time on doing something new with CCS.

[sandalcandal]

So if I see a regular diode FULL BRIDGE rectifier (oh wrong channel, sorry!) then it should be right?

A bang-bang square wave inverter with no 0V dwell time might (or might not) cause some issues for a synchronous rectifier but a passive diode rectifier would also likely see some increased loss due to hard switching caused fast reversal of voltage but unlikely to cause malfunction.

Some teardowns of other OBCs for reference:

(weird semi-active bridge?)


(diode rectifier in Tesla Model 3)

Another suggestion, maybe see if there are any scrap/salvage OBCs for your EV you can get before you go trying to tear into your own working car.

However, if efficiency is the concern, current state of the art gets over 99% efficiency with "pure sine" output using WBG (wide band gap) semiconductors and a totem pole topology e.g. https://training.ti.com/demo-900-v-5kw-bidirectional-ac-dc-converter-gan Obviously there's a bit of a cost premium but if you're making something new, might as well.

The thing is, apart from the cost and weight, I have to factor in the OBC inefficiency. And the battery efficiency (intra cycle hopefully), then the MPPT efficiency. It doesn't just add up, it multiplies out to a pretty bad deal.

Yeah, you could get better solar battery to EV battery efficiency if you get a DC-DC system working and bypass the OBC I guess. Not sure the cost and effort would be justified without seeing your actual numbers however.  I'd expect an OBC to have >90% efficiency and a well implemented DC-DC to have ~95% [98% on the high end]. 5% improvement might not really justify the costs. ~10% in a "best case" might also not be worth it if the costs involved are too high.

[FyKnight]

A bang-bang square wave inverter with no 0V dwell time might (or might not) cause some issues for a synchronous rectifier but a passive diode rectifier would also likely see some increased loss due to hard switching caused fast reversal of voltage but unlikely to cause malfunction.

I'd worried about this and will probably put in some very small zero time because I'll be scared to have both paths on at the same time with differing on- and off-transition times. Doubt it'd make it much better for the rectifier though. But then again unless there's some other capacitance in the panels (probably insufficient given the speed of MPPT scanning algos) I'll lose efficiency whenever I'm not passing that current through. And now I want to add some capacitor or inductor haha... but no, I really don't want to end up designing most of a regular inverter here! Let's see how I go with the dumb approach first.

Another suggestion, maybe see if there are any scrap/salvage OBCs for your EV you can get before you go trying to tear into your own working car.

Hahah yes, and if I can get a spare battery with it that'd be even better! You read my mind and I was looking on Pickles last night. No-one has crashed their shiny new MG EV recently though.

Yeah, you could get better solar battery to EV battery efficiency if you get a DC-DC system working and bypass the OBC I guess. Not sure the cost and effort would be justified without seeing your actual numbers however.  I'd expect an OBC to have >90% efficiency and a well implemented DC-DC to have ~95% [98% on the high end]. 5% improvement might not really justify the costs. ~10% in a "best case" might also not be worth it if the costs involved are too high.

You're definitely right, though in my EV driving experience, 10% extra would be really nice sometimes! If you are limited in solar array space then it's even more important. And if you also build in the MPPT controller to avoid extra conversion stages and battery losses then you're even further ahead.

I wasn't sure if I could link YT videos from other channels here but I see you did (and thank you for those teardowns!). Here's the MPPT + DC-DC converter board that is on its way to me now, not powerful enough for my application, but let's see how it performs:

Edit1: timecode in link didn't work: go to 21:15 to see the board
Edit2: OMG I just realised the source of one of those videos. I have so much to learn. Thanks again everyone!

[sandalcandal]

I'd worried about this and will probably put in some very small zero time because I'll be scared to have both paths on at the same time with differing on- and off-transition times. Doubt it'd make it much better for the rectifier though. But then again unless there's some other capacitance in the panels (probably insufficient given the speed of MPPT scanning algos) I'll lose efficiency whenever I'm not passing that current through. And now I want to add some capacitor or inductor haha... but no, I really don't want to end up designing most of a regular inverter here! Let's see how I go with the dumb approach first.

"Dead time" is the technical word used in power electronics for this period when all the switches are turned off. You definitely need some sort of buffering (DC-link) for smoothing power off the panels. Photovoltaics don't care about the average power, they need the instantaneous power draw to be at their maximum power point. A capacitor would generally be required to get the high-frequency smoothing required but I don't have the expertise in solar power systems to know more niche methods.

You're definitely right, though in my EV driving experience, 10% extra would be really nice sometimes! If you are limited in solar array space then it's even more important. And if you also build in the MPPT controller to avoid extra conversion stages and battery losses then you're even further ahead.

Yes, if you are limited in space for your solar array then squeezing the most out of the available solar area is important, you can't just redirect lowered power electronics costs to buying more (or to an extent better) panels in the limited solar array case.

[FyKnight]

I finally took the plunge and undid the excessive number of screws to open the OBC.



Here's my interpretation (any of this could be wrong!) The top left is the Type 2 AC cable coming in. The top right had better be the battery — didn't actually test that though.

The grey pads are some heat transfer material, and luckily half of it stuck to the lid so I could see what was under them better There's a nice pair of inductors which I expected. Some absolutely massive FETs pinned against the wall. A capacitor bank under the black plastic at the bottom right, more boxy capacitors dotted around.

In the middle of the board (actually poking up through a hole in a mezzanine board) are what looks to be two good size transformers. Maybe some sense transformers above them too.

So yeah, given the proliferation of probable transformers, I'm going to guess direct DC into it is a bad idea. Please correct me if I'm wrong though. I have a higher resolution photo (and videos) too if helpful.

[f4eru]

DC can make a PFC go mad...
DC can get an arc over a relay contact keep burning and not extinguish.
DC can mess up active X cap discharge circuits.

I personally would not take the risk for an OBC, but I would for a smaller PSU.

[uer166]

The inductors on bottom left are the PFC boost, and you can see the capacitor bank on bottom right, it then loops back around to upper right through transformers. It's probably some sort of LLC/PSFB/DAB converter. There is nothing fundamentally preventing the use of DC input here, but as was mentioned, you may explode the input contractor that's not rated for DC (fatal), or confuse the PFC controller. I'd recommend trying it just so we see the result for funsies, assuming it wouldn't be a huge loss! 

[FyKnight]

There is nothing fundamentally preventing the use of DC input here, but as was mentioned, you may explode the input contractor that's not rated for DC (fatal), or confuse the PFC controller. I'd recommend trying it just so we see the result for funsies, assuming it wouldn't be a huge loss!

Excellent, maybe I can replace that contactor with one that is rated for DC then. I'm willing to risk confusing the PFC controller. Technically, those pins could expect to see DC in some (never implemented AFAIK) charging systems so... it's not inconceivable that the OBC would be safe to receive it even if it didn't charge with it.


Yeah breaking this OBC would be a big loss, being the only way to charge my regular car. I reckon I'll try it with square wave input first, and wait to pick up a wreck to try actual DC.

Thanks very much for your replies folks!

[Yorch]

Great thing! Since years i have this idea to try this but did not dare yet.

Any news from your side?

Actually i have a Hyundai Ioniq Facelift. At the beginning it had a failure in the DCDC converter, causing the 12V Battery to go empty. Because the DCDC is withing the OBC, they changed the complete unit, but send it back to Korea, so no chance for me to teardown. On the Unit its written Input Voltage 85...285V, 20...100Hz.
These 20...100Hz makes me belive that frequency in fact does not matter at all....


Your point that it has to be "DC-Mid" save is worth thinking. The question is, when designing a OBC, do i have to consider it will be connected to failed DC-Mid DC-Charger? Since now only Tesla uses this at its old superchargers, but they will turn on the voltage only if a Tesla has identified by the CAN-Bus comminication over the pilot pins...

[f4eru]

There might be a misconception here.
The DC-MID mode is not passing the power through a converter. it is contacted directly to the battery voltage, and the converter is in the charge point.

[FyKnight]

Any news from your side?

I bought all the equipment to do it but didn't get around to trying it yet, mainly due to being distracted with another project (building my new workshop!)

However, I've since heard separately from a contact who has tried it with 160V DC from batteries directly into the AC pins, and not only did it not break the OBC, it successfully charged his Volt.

There might be a misconception here.
The DC-MID mode is not passing the power through a converter. it is contacted directly to the battery voltage, and the converter is in the charge point.

The point is not to use the mode (which is rarely supported), but that its existence means the OBC could be designed to safely accept DC on those pins, and not melt a coil or lock a contactor. So it might be safe to test it even if it didn't work to charge.

[Seekonk]

I have a solar powered summer home and am thinking about getting an EV for that location.  Opposite reason from most, I want to bring power home so I can run an AC.  I do a lot of direct power from the panels and have a very minimal battery. A shunt regulator to a water heater keeps the array voltage stable. I run a LG washer right from the panels using an old MSW inverter with no battery. Just feed HV panel DC right into the H bridge and some 12V to run the control electronics.  The panels will supply the current limit.  The H bridge of the inverter will keep running regardless of the voltage and not cause any faults.

[f4eru]

The point is not to use the mode (which is rarely supported), but that its existence means the OBC could be designed to safely accept DC on those pins, and not melt a coil or lock a contactor. So it might be safe to test it even if it didn't work to charge.

it's not. DC is only present if negociated sucessfully.
DC will probably charge your car. But it remains basically unsafe.

[TimNJ]

Just generally speaking, if there is a fault which should cause a fuse to blow, the AC rated fuses may not be able to interrupt DC. DC is pretty hard to interrupt by comparison, so you should always check the specification of the included fuse(s) and check if they have a DC rating (with ample breaking capacity). If not, you should consider adding a proper DC rated fuse, with enough breaking capacity for the prospective fault current of the installation.

[FyKnight]

I run a LG washer right from the panels using an old MSW inverter with no battery. Just feed HV panel DC right into the H bridge and some 12V to run the control electronics.  The panels will supply the current limit.  The H bridge of the inverter will keep running regardless of the voltage and not cause any faults.

Nice work, that's a great idea! I'll do the same for my test now (already have the MSW inverter) instead of building my own.... thanks!

it's not. DC is only present if negociated sucessfully.

So is AC but I doubt it would physically damage it to have AC present outside the negotiated state.

DC will probably charge your car. But it remains basically unsafe.

For sure, yeah look I hear you and I realise this is optimistic thinking. But if I break something then it'll be a fun project trying to fix it

Just generally speaking, if there is a fault which should cause a fuse to blow, the AC rated fuses may not be able to interrupt DC. DC is pretty hard to interrupt by comparison, so you should always check the specification of the included fuse(s) and check if they have a DC rating (with ample breaking capacity). If not, you should consider adding a proper DC rated fuse, with enough breaking capacity for the prospective fault current of the installation.

Thanks for the suggestion! Thinking about it, this should only be an issue on the input side right? After that it'll all be the same inside the OBC. However coming from PV panels their Isc is only about 10% higher than Imp, so not much for a fuse to do there. OK well sure, if I had a bigger array than the OBC could use (over ~7 kW / 32 A Isc) then this would be possible... that'd be a nice problem to have

[richard.cs]

it's not. DC is only present if negociated sucessfully.

So is AC but I doubt it would physically damage it to have AC present outside the negotiated state.

Kind of a yes but, I am aware of at least one AC implementation that is damaged by a corner case. On a Renault Zoe single phase charging is achieved by connecting L3 to N with a contactor because the charger only has inputs for L1 L2 L3. The trigger for this was (or used to be) that there was no voltage on the DC bus, but L3-N voltage was not checked before closing the contactor. Fine most of the time, but consider if L3 does have volts relative to N, but L1 and L2 are open circuit or misconnected to the same phase as L3. Then there is no DC bus voltage, but closing the contactor shorts L3 and can result in it welding closed. I think it has been fixed in newer models.

Another one, on cars where 3 phase charging is 3 separate single phase chargers it is quite possible that connecting all three phases to the same phase may result in triple neutral current and a very unhappy cable.

Not all cases that might reasonably be expected seem to have been considered by designers.

[TimNJ]

Just generally speaking, if there is a fault which should cause a fuse to blow, the AC rated fuses may not be able to interrupt DC. DC is pretty hard to interrupt by comparison, so you should always check the specification of the included fuse(s) and check if they have a DC rating (with ample breaking capacity). If not, you should consider adding a proper DC rated fuse, with enough breaking capacity for the prospective fault current of the installation.

Thanks for the suggestion! Thinking about it, this should only be an issue on the input side right? After that it'll all be the same inside the OBC. However coming from PV panels their Isc is only about 10% higher than Imp, so not much for a fuse to do there. OK well sure, if I had a bigger array than the OBC could use (over ~7 kW / 32 A Isc) then this would be possible... that'd be a nice problem to have

To be honest, I did not read this whole thread in great detail, but it was just a general piece of advise. I'd guess that a large percentage of (switchmode) AC-DC converters, which work by rectifying AC first, are compatible with DC input. Maybe there are a handful of controllers/designs which will freak out if they don't detect a zero-crossing. Hard to know. If you test it and it works then...I guess it works?

And to answer your question, yes, only matters on the input side...otherwise it's all the same internally. My point was: Whatever AC rated fuse was originally specified/included in the equipment is not necessarily capable of breaking DC, if a fault situation arose. Same applies for circuit breakers, contactors, and switches.