HT3525+Lyen124115+30S want more power... safely

[hjns]

Hi all,

I added some solder to the shunt in my Lyen MKII 12 FET 1445 controller as described here. I also re-calibrated the CA as described here using 9.0 and 10.0A current from my iCharger in foam-cut mode over the Ground-Phase circuit. It showed that I had to decrease the shunt value to 1.645 mOhm (instead of the 2.000 mOhm). I also programmed the controller as described in Methods^thread to 58A/145A current with block time of 3 seconds. You may remember I have a HT3525 with halls and 30S 126V lipo. When driving up a 10% hill, both motor and controller get hand-warm after 500m, and I see peaks amps over 60A (7.5kW). Outside temp is around 65 degrees Fahrenheit.

How far do you think I can push it? I do not want to start drilling holes yet... and I will add a temp sensor to both controller and motor. But for the next steps, should I first add some more solder to the shunt to change the shunt resistance to below 0.500 mOhm and re-calibrate the CA? Or should I first stick with increasing current in the programming?

Thanks!

KR,

[edit: title edited to reflect correct MOSFETs]

[ZOMGVTEK]

At 7.5kW peak, you are allready above what I consider to be the 'safe' limit for your hardware.

If you want more power, you need a new controller, serious motor cooling, or a bigger motor.

[zombiess]

You mean you are running 60A through a 12 FET 4115 controller with 3.0S block time and 145A phase? It's on borrowed time especially if you are running it in anything bigger than a 20" diameter tire. Just read Binlagins thread about blowing up his 18 FET 4115 controller, worked great for a week on his greedy power settings then popped some FETs.

On my high modded 18 FET Lyen MK2 4115 controller we turned it down to 60A battery current 100A phase (7.5KW continuous use) for racing this weekend at Grange (mainly to allow better throttle control but also to ensure the controller didn't heat up too much). I have temp sensors in it and after 12 laps the FETs were sitting about 65C with the internal case temp about the same. One of the first things I do is turn block time down to 0.1 Sec (need to hex edit the software to do this) to help keep it alive. You are already well beyond the safe limits of that controller. You really should have an 18 FET to try and run 60A continuous at 145A phase and even that might be a bit high. A 24 FET 4115 would be good to go but you also risk unmatched FETs the more you put in parallel, however most have had good luck with the 24 FET controllers. I've gone to building my own controller for high voltage setups to make sure everything is exactly as I want, but it can be tedious with the mods I do.

I'd suggest dropping your block time down to 1.0S, don't hammer it from a dead stop and drop your battery current down to 45A and phase amps to 125A if you'd like to see it live a while. You might get lucky and be OK, but your certainly pushing the controller really hard right now.

[hjns]

Hi Zom and ZOM,

Thanks for your replies. Yes, I mean 145A phase and 60A battery. OK, I got the message not to push it any further.
I found the hex edited file (Gensem, thanks!), and will take protective measurements. Temporary bursts of 7.5W is mighty fun though, and from what I read the only reason that my controller hasn't blown up is that I don' do 7.5W continuously.

Thanks again!

KR

[binlagin]

Yeah, I'd defintly listen to what Zombiess has to say about this.

I ended up destorying a FET when running 60amp battery and 120 phase amps with a stock 18-FET Lyen controller. I'm seeing alot of people doing this with their un-modified smaller controllers... but it's only a mater of time until they go POOF!

I'm going to be preforming the mosfet tie together MOD that Zombies has documented on a rainy day in the coming weeks. I just have to decided if I'm going to switch to the Mica insulator. I also plan on adding some form of ventilation

[zombiess]

Yeah, I'd defintly listen to what Zombiess has to say about this.

I ended up destorying a FET when running 60amp battery and 120 phase amps with a stock 18-FET Lyen controller. I'm seeing alot of people doing this with their un-modified smaller controllers... but it's only a mater of time until they go POOF!

I'm going to be preforming the mosfet tie together MOD that Zombies has documented on a rainy day in the coming weeks. I just have to decided if I'm going to switch to the Mica insulator. I also plan on adding some form of ventilation

Don't use mica insulators, they are a PITA. Use these http://parts.digikey.com/1/parts/988564-thermal-pad-220-007-gray-173-7-220p.html You can trim them with scissors, much much easier. If you reuse the stock screws just make sure you have no shorts when you are done. That's why I switched to 6-32 nylon screws. 6-32 because they are bigger than the 3mm stock ones which means I can tighten them up more than a 3mm version which I couldn't find here. I kept having shorting issues with the stock insulators because the extra depth the buss bar eyelets add allows the screw to contact the FET tab. They do make longer insulators for the screws but I wasn't going to wait for them to come in the mail so I decided to give nylon screws a shot. Been working great since and haven't had a single issue, just don't over tighten them. I use bronze lock washers, feel them squish and the resistance increase then go 1/4 to 1/2 turn further and call it good.

If you use nylon, sacrifice 1 or 2 so you know how far you can go before they break and can get a good feel for them.

[hjns]

I think I have found my next projects:

1. Oil cooling of my motorhub: many nice threads around; Reid Welch, Dave b, andynogo, bigmoose (very helpful), itchynackers.
2. Cooling mods of my controller. I may end up going the Zombiess way...

By the way, I don't mind replacing FETs anymore. However, I still have a deep fear for destroying the other components on the board....

Anybody thought of oil cooling of a controller?

[hjns]

I sent Methods a PM with the same question. His response was to max out the software. He would not hesitate to run 100V 120A on an 18fet especially with a low KV motor (like my HT3525). The real question is how long I can keep it in the current limit before back EMF starts to drop. I will try and keep below 60A, so as not to hit the wall too often. Methods also told me to worry less about the long term ride and focus more on the instantaneous efficiency. "One hard acceleration off the line (after getting the motor heat soaked) can cause failure... but if you keep a running log in your head of how hard you have been hitting it you can pretty much do whatever you want."

In other words, I keep below 40A for the long term ride, and I carefully accelerate up to 60A once in a while. I brought block time back to 1.0 second. From what I have been reading on oil cooling, it seems like the best thing to do to keep temps below 60oC for both motor and controller. I read about a PC soaked in oil. Will be fun to try that with a controller....

[Lebowski]

Anybody thought of oil cooling of a controller?

Yes actually. But I've come to the conclusion that a lot more is possible with a limited amount of FET's than
what people think. You don't need 36 FET's for 10 kW...

[hjns]

Yes actually.

Did you try it?
Anyway, this morning I rode to work and had peak Amps of 80A (80x126=10kW) with max speed of 82km/h (50mph). It was only for a moment, and neither controller nor motor were very warm, because I was holding back for most of the commute, averaging at 40km/h.

[Lebowski]

Yes actually.

Did you try it?

No. But I would get a slid aluminium bar, dril a hole lengthwise for the oil to flow through and screw the FETs to this.

But 6 4110's at 80V and 70A battery (100A phase, about 10W of dissipation per FET) would give
you a 5.6 kW controller. You don't need a huge amount of FETs for this power level.

[hjns]

No. But I would get a slid aluminium bar, dril a hole lengthwise for the oil to flow through and screw the FETs to this.

But 6 4110's at 80V and 70A battery (100A phase, about 10W of dissipation per FET) would give
you a 5.6 kW controller. You don't need a huge amount of FETs for this power level.

Hmm. I was thinking more in line of filling the whole controller with oil, with a small pump and a small radiator to the front of the bike. My 12 FET 1445 controller seems to be able to deal with 126V 80A max peaks. I read somewhere that these 1445 FETs can take 74A, so I am already above te rated limits.

[zombiess]

Are you talking about IRFB4115 FETs or is this some new FET being used? I think you mean IRFB4115s.

[hjns]

Ooops, sorry about that.

I have this Lyen controller: viewtopic.php?f=31&t=18214&hilit=4115
It carries 12 of the IRFB4115 MOSFETs.

[hjns]

I have been thinking about these IRFB 4228 MOSFETs:
almost the same specs as the 4115, but higher peak current capability (170A instead of 104A), similar resistance (12mOhm vs 11mOhm), and similar form factor (TO-220). Would these take more high current abuse?

link

[grindz145]

Impedance is really all that matters. I'm confused as to how you can have a higher impedance and a higher maximum current capability. Doesn't make sense to me...

[ZOMGVTEK]

I have about a thousand miles on my little 12 FET 4110 Infenion @ 8kW. I plan on running it at 9-10kW when the 5404 build is done.

The only real reason why you need more FET's for 10kW, is for the bigger case to sink additional heat. My controller is fine after a few WOT runs. Just make sure its getting blasted with air and it never gets above warm. Even lugging up a hill at slow speeds and 6-8kW for 30+seconds, it never gets past 'hot'. I'd be comfortable going WOT as long as my controller isn't burning hot. A friend once had a 6 FET that was FAR too hot to touch, and even smelled. It still works fine and runs burning hot sealed in a bag. But thats a 22V bike with 60V FET's at 500-550W.

The issue here, is the 4115's are fairly high Rds(on) as compared to the 4110's, as a result they take less power. Keep it below 100V, and a 12 FET 4110 will do 8kW no problem, 10kW if you really want to push it or have well matched FET's. The way I see it, replacing 12 FET's isnt a big deal, so i'm willing to run my controller near the edge. I carry a spare 6 FET on the bike to limp me home at 3kW just in case.

For some reason, nobody really tends to push bigger controllers. I see people cook 6/12/15 FET controllers from way too much power, but most failures for the 18 FET and up are not purely 'power' related.

[zombiess]

I have been thinking about these IRFB 4228 MOSFETs:
almost the same specs as the 4115, but higher peak current capability (170A instead of 104A), similar resistance (12mOhm vs 11mOhm), and similar form factor (TO-220). Would these take more high current abuse?

link

You are reading the spec sheet wrong.

IRFB4115@25C = 104A and @100C=74A
IRFB4228@25C = 83A and @100C = 59A

[hjns]

... I guess I just showed my ignorance in this field....

Anyway, the main reason to get the low kv HT was to get up the hills quickly. That it does. Then the main reason to get the Lyen controller with the IRFB4115 mosfets was to be able to run a high voltage so that I can do some speed as well. That it does too. Now, I understand that the milion dollar question is how to get both combined. From what I read, Methods and you prefer the 100V max route using the IRFB4110 fets. That makes 10kW at 100A battery current, and you can do that for longer periods of time? With the 4115 fets and 126V I "only" need 80A to get to 10kW but that is peaks only. Now I understand that there is a major difference between sustained currents and peak currents, and that my 4115 fets won't be able to sustain these currents, where the 4110 fets will. Am I correct in these assumptions?

[zombiess]

I'm not a fan of running anything less than an 18 FET IRFB4115 controller. Even on the 18 FET version I believe Lyen recommends only running 45A max and I personally think 60A is about the max one can take without modifying it (and even that could be pushing it).

It's ultimately your choice on where you run your controller. If you don't mind pushing the boundaries and frying a few FETs in the process, go for it. I tend to stay on the safer side myself with close monitoring to let me know when I'm getting close to the limits. The greed for power will kill parts, just be ready to pay.

It looks like with your motor you only need 50-60A to be into the BEMF area @ 126Vwhere amps will drop. Why not run an 18 or 24 FET and not have to worry so much?

Also, be careful with that pack voltage, you are in the lethal if touched wrong range, treat it with respect.

[ZOMGVTEK]

The thing that determines the 'peak' power is going to be the MOSET's themselves. They have a somewhat 'soft' limit. It's not like voltage, where even going 1V over might be enough for the FET to go short. You can go a few A's over and things should be OK if you keep them cool and the duration short enough.

Continuous power handling is an entirely different beast. It's really only determined by how much heat the controller case can dump. If you water cool a 18FET and do a few mods to get a really low delta between the silicon in the FET and the heatsink, you can basically run the controller at its peak power 24/7. Otherwise, continuous power can be as low as 10% of 'peak' power.

According to this spec sheet...
http://www.irf.com/product-info/datashe ... 110pbf.pdf

The IRFB4110PBF TO-220's commonly used in high power e-bike controllers should be good for 180A at the silicon level, which is the more important number if you really are interested in pushing things, especially since we don't tend to be riding the current limit that long so long as it is sufficiently high. A good assumption is that for every time you double the number of FET's, you can handle 1.5x the current. This means every additional FET adds roughly 2/3 of its max rating. Since the controller only really cares about phase current, we will use that, assuming it's 2.6x higher than the battery current, at least in the lower speed region thats more critical. For an 18 FET controller, we have 3 phases, making 6 FET's per phase, which is then split up evenly into a high and low side. This means your power handling calculations should be done with the assumption of 3 FET's doing the 'work'. With 4110's, you can handle a max of 180 + 2(2/3)180 = 420A in a 18 FET controller. This is roughly 160A battery current. Setting the battery/phase to 160/420 respectively would roughly be asking the controller for 100%, which tends to be a bad idea, since this limit can often be loosely enforced at times. Playing it 'safe' would put the max limit around 20% lower than this, 125/200. This should be OK if you can keep the controller cool.

100% limits are roughly 16kW for an 18 FET 4110 @ 100V.

Moving on to the 4115's....
http://www.irf.com/product-info/datashe ... 115pbf.pdf

104A silicon limited. Same 18 FET controller, 104 + 2(2/3)104 = 242A phase, 93A battery. Assuming you run 150v into these (bad idea), you can get roughly 14kW out of a 18 FET IRFB4115PFB controller, absolute peak. Take 20% off this to 'play it safe' and you are down to just over 11kW. Even trying to get 11kW out of one of these would be a struggle, but it should be possible with the right tweaks.

100% limits are roughly 14kW for an 18 FET 4115 @ 150V.

Run the 4115 at 120-130V and the limitations are even more obvious. The IRFB4110PBF's are just better FET's for power. Unfortunately >100V right now isnt the best way to get a ton of power, at least with off the shelf hardware.

I should probably note, these 100% figures are not reasonably obtainable. Running a controller even at 80% of its limits takes a mad man.

[zombiess]

ZOMGVTEK, if you read note 1 of the spec sheet you'll see it's limited to 120A due to the wire bond limit. The legs on the FETs really prevent sustained currents higher than about 75A, but we are operating them in pulse mode so we can get to the 120A limit. Care must still be taken at low speed when you run a controller on the edge as well since the FETs are on for a longer period of time due to the commutation time of the motor being longer. This is also why having a high block time setting is bad, I always change it to 0.1S.

Don't forget the high side FET bank has switching losses + resistive losses. The low side bank only has resistive losses since it's doesn't see PWM. You should never calculate off the 25C ratings, always assume the worst and calculate off the 100C ratings. In the IRFB4110's case these are both 120A, but the resistive losses are about 1.7x higher at 100C for the IRFB4110 bringing each one from 4.5mOhm to 7.6mOhms each.


hjns: if you try running 80A on your 12 FET IRFB4115 controller it will probably work for a while, but if it fails don't be surprised.

Cliff notes:
Buy a bigger controller than you need. You can also custom build or improve your controller like I do and push the limits farther. My 18 FET controller has been run over 12KW many times at 125V, but the key is it's not stock and I don't stay at high current very long due to running small wheels/tires so I get up to speed fast and the amps drop.

[ZOMGVTEK]

zombiess, I was just going for the 'absolute best case senario' type of limit, assuming any possible mods are in place to allow the controller to run near its limit. This also includes the assumption that the FET's are reasonably matched, and this peak power isnt sustained for more than a few seconds. Take this number as a general guide, and cut 20% off, then you have a fairly accurate 'REALLY PUSHING IT' limit. If you have a shoveled together controller in stock form with FET's placed in the order they were in the tube, cut this limit in half.

I did some destructive testing on IRFB4110PBF's and they take >120A for quite a while. I didn't have enough heat sink to run it for hours at 120A, but it will probably take it. I didn't have enough supply at the time to run past 145A or so, but I couldn't get the bond wires to melt. Obviously, being that this is a single FET, its going to share current quite well. Thats where the 1/3 loss factors in for more than one FET. I feel comfortable saying that IRFB4110PBF's can sustain the 100A or so I claim.

And yes, I would agree a 18 FET is the minimum for the 4115's. Plus, 3 FET's can share current fairly effectively, at least compared to really big banks in a row.

[zombiess]

I did some destructive testing on IRFB4110PBF's and they take >120A for quite a while. I didn't have enough heat sink to run it for hours at 120A, but it will probably take it. I didn't have enough supply at the time to run past 145A or so, but I couldn't get the bond wires to melt. Obviously, being that this is a single FET, its going to share current quite well. Thats where the 1/3 loss factors in for more than one FET. I feel comfortable saying that IRFB4110PBF's can sustain the 100A or so I claim.

That's the first I've heard of someone doing this. This little tid bit of info is being stored in the memory banks (hopefully they won't get corrupted). Thanks for providing this info. It really seems the International Rectifier under specs their components to ensure they exceed their data sheets as has been said by others on here. I don't know about you, but personally I think the IRFB4115 FETs get a bad wrap and I do think I could build a 12FET controller that could sustain a 80amp battery draw. Sounds like a fun game and when I have some free time I might just give it a shot since I still have about 100 IRFB4115s left and a few 12 FET controllers. I'm actually finishing up a 12 FET IRFP4568 controller add on board I made and started another 36 FET IRFB4115 controller in a smaller case so I can mount it where my 18FET unit is now. I'm going to have more high power high voltage controllers than I know what to do with soon LOL, going to have to sell some off on here.

How long did you run the test for on that IRFB4110 FET? Seconds, minutes? Even just a few seconds is a killer test... ok, they lived so I guess it wasn't literally a killer test, just a torture test

I assume that when I parallel FETs randomly I'm going to only get 66% of the paralleled current rating, but when they are closely matched for turn on voltage in big banks such as a 24 or 36 FET where it's the most critical I think it could be as high as 90% (just a guess, several factors in play such as switching frequency can come into play). I think I'm pretty lucky with my 18 FET since it's just random FETs installed by Lyen. They are most likely from the same batch (which improves the chances of them being close) but that doesn't mean each bank of 3 has matched turn on times based on my order of 100 of them and then grouping them by measuring resistance at 4.0V Vgs. Got quite a resistance spread.

[zombiess]

hjns: sorry if ZOMGVTEK and I are getting a bit techy and off topic on you, but it's related to your initial question.

Looking at the simulator I'd say go for the 18 FET controller IRFB4115 controller and set it for 60A battery and probably 140A phase and you'll probably be OK unless you need to climb a bunch of hills. Then go for the 24 FET version which Lyen says should be fine at anything under 100A but I'm not sure what phase current setting, personally I'd probably start at 80A battery and 150a phase and go from there on 24 FET IRFB4115 from Lyen.

Personally I'd just stick with a 12 FET IRFB4110 or 18 FET IRFB4110 and run 24S (100V) on that motor and live with it only going a top speed of 40MPH/65KPH but be reliable and able to climb well without worrying about the controller, not to mention fast acceleration which I find more fun than top speed. ZOMGVTEK, what do you think?

hjns, what is your comfort level with working with electronics? Think you could attempt the same mods I've done (I copied markcycle so it's not my mod idea, I'm just impressed with the results from it)?