Am I missing something? (Long distance power transfer via transformers)

1st Official Post

    So I was messing around with some long distance power transfer today, and ran into a bit of a snag. I was getting far, far less energy at the other end than I was expecting, so I set up a small system to test what's going on. The rig I used was batbox - LV redstoned - MV redstoned - HV redstoned - 1 piece of 4x iron cable (which should incur no loss), - HV - MV - LV - batbox. I stuck a rechargable battery into it, and the initial batbox has 16 EU left in it, but the ending one only has 7808 EU. That's 2176 unaccounted for. I thought maybe some packets got caught in the transformers, but the maximum that could have gotten caught would be 3 pulses of 512 in the HV, 3 pulses of 128 in the MV, and 3 pulses of 32 in the LV. That only adds up to 2016...leaving 160 EU missing.


    I then, through a lot of patience, set up a batbox with exactly 2048 EU in it. I wired it, knowing that if all is well, there should be absolutely no packet waste whatsoever, the end batbox should end up with 2048 EU in it. It ended up with only 1664.


    Screenshot of the setup is here:


    So am I missing something? When I put 2048 EU into the batbox, it should have pulled packets of 32 to hit 128, then pulled packets of 128 to hit 512, then packets of 512 to hit 2048, then transferred a single packet of 2048 to be broken down into 4 pulses of 512, 16 pulses of 128, and 64 pulses of 32, right? I don't notice any loss when sending long distance transmission stepping up from HV to EV then back to HV using an MFSU on both ends, but for sending small amounts of energy like what you'd get from a wind farm two MFSUs seems extreme...you might as well just use glass fiber at that point. Even MFEs aren't really cost effective in that regard.


    Anyone have any advice? Am I just being a noob and missing something obvious?

    How does transofmer work when transforming from higher to lower current ? Is there possibility that it needs full high packet to turn to 4 small packets ? If yes, it may answer the question, since some energy may be left in other transformers aswell. I was playing with wiring and i came to conclusion that no EU is lost during transport, however it gets divided into transformers. I was doing so by placing redstone dust into input MFSU and checking output. If there was some way to measure stored energy in transformers, i would be completely sure about that.

    That's why I tested it with 2048 EU in the first batbox. The upstepped LV transformer should have pulled 32+32+32+32 to make 128, the upstepped MV should have pulled 4 of those packets of 128 to make 512, and the upstepped HV should have pulled 4 of those 512 packets to make 2048, then the HV should have downstepped that 2048 packet to 4 512s, and so on. There should have been absolutely nothing left over, but a large chunk appears to have just vanished.

    Transformer GUI should be added to remove all doubt. You can never be sure something bad happened in TFs that made them somehow not correctly divide/compress packets. I believe there is no energy loss in setup you presented. I will try it myself later and post results.

    Quote from Nightdagger

    That's why I tested it with 2048 EU in the first batbox. The upstepped LV transformer should have pulled 32+32+32+32 to make 128, the upstepped MV should have pulled 4 of those packets of 128 to make 512, and the upstepped HV should have pulled 4 of those 512 packets to make 2048, then the HV should have downstepped that 2048 packet to 4 512s, and so on. There should have been absolutely nothing left over, but a large chunk appears to have just vanished.


    probably just something stuck in the buffers. I've got a EV system for wind power generation and it transmits to the ground fine, only losing 3-5 or so during transmission.

    Quote from FnordMan

    probably just something stuck in the buffers. I've got a EV system for wind power generation and it transmits to the ground fine, only losing 3-5 or so during transmission.

    How long is your system ? How much EU/t your farm can produce ? Losing 3-5 out of how much ?

    I'm running a much longer EV system; with around 32 eU/p lost out of 2048eU/p the ~1.5% loss rate of 3 rubber insulated HV cable really beats the cost of rewiring it with glass (esp since it's temporary). Now, for powering a remote quad mining rig I'm finding that a string of LVTFs and copper wire quite managable; though in the next mining phase I think I'll have the power strip inside, rather than along, the ceiling. That got annoying fast.

    Quote from MJEvans

    I'm running a much longer EV system; with around 32 eU/p lost out of 2048eU/p the ~1.5% loss rate of 3 rubber insulated HV cable really beats the cost of rewiring it with glass (esp since it's temporary). Now, for powering a remote quad mining rig I'm finding that a string of LVTFs and copper wire quite managable; though in the next mining phase I think I'll have the power strip inside, rather than along, the ceiling. That got annoying fast.

    Glass fibre is there for short HV wires and perfectionists. But still, 1.5% of 100M EU is 1,5M EU which may sometimes be just enough to spare 10 diamonds to make that damn glass fibre. Depends on how much you are going to use that particular line.

    I think the problem is just that I'm not testing with a large enough sample size. After all, some incompleted packets could be getting hung up in the transformers, so I decided to test it with much larger amounts to minimize that. I set up the same system as last night, except with an MFSU on the receiving end, and fired 300k energy at it through 300 single use batteries in the batbox, and ended up with 298624 EU in the receiving MFSU, which is only about 0.5% loss. The missing energy is only 1376 EU, which could easily just be hung up in the transformers waiting for a bit more energy to come at it to pulse.


    Just for the sake of completion, I then wired the same system, but with 40 pieces of 4xHV cable instead of just 1, to incur some loss in the system, and pumped 300k through the batbox. I ended up with 293888 (97.9% efficiency, or 2.1% loss), getting packets of 2016/2048, only 1.5% loss over a 40 block span...the extra loss I attribute to hung packets. Another 300k through the same batbox left me with 588288 in the receiving MFSU...only 2% loss, including the hung packets.


    I'm thinking my initial test was flawed in that I might have had the batbox hooked up to the transformers when I started feeding energy into it to hit the 2048 and disconnected it somewhere along the way. Even though the receiving batbox wasn't hooked up, the transformers might have pulled a few packets to start filling their internal buffers. Regardless, for transmitting small amounts of energy, EV is inefficient as hell, but for large amounts it's definitely a great option, and the more energy you transmit through it, the less you'll notice the hung packet effect.


    So yeah. I was missing something. More power.

    Quite agree on the quantity transferred. Glass fibre is for reactor grade connections and primary backbones. The windgen tower will at most produce between 128 (worse case for 8 gens) and 256 eU/t, so packing it up in to pulses... actually I'll probably go through the effort of rebuilding it with copper and just stuff an MFE at the top for buffering. However that's later, after I get a working breeder reactor and build up much closer to it so that it makes sense to better integrate it in to the rest of my infrastructure. Right now it's powering miners and overflowing to a Massfab...


    As for the 'hung up' energy. Just make sure your input is close to a multiple of 2048. If you use a full batbox (40K IIRC?) the 'hung up' portion would be... 1088 (40000%2048) however there are also other factors, like storage units secretly hoarding a bit of 'excess' charge and not displaying it. If you want to be sure, fill it with full batteries alone.

    Actually, what I was looking at it for is powering a bank of geothermals that I don't want cluttering up my house. I want to put the power plant a fairly decent ways away from my house so I'm not constantly tripping over wires. I'm planning on 10 geothermals in the bank, but I don't want to splurge on an extra MFSU just to act as a storage buffer prior to transfer and a batbox or MFE as a buffer wouldn't be able to transfer the power it'll be taking in, so I'm going to wire all of the geothermals directly to the upstepping LV transformer. That way I don't have to worry about the buffer, because the transformer itself will buffer the signal and send a packet whenever it's full, and I can save on resources while still getting a good chunk of the power out of the power plant, as well as reducing clutter in my house. Which is worth it's weight in gold to me...or iron as it were...but not diamonds :D

    Quote from raGan

    How long is your system ? How much EU/t your farm can produce ? Losing 3-5 out of how much ?


    right now it's making ~75EU/t at the end reciving point. The source is 20 wind mills and 8 solar panels* on the top of the world.
    They go through a LV, MV and HV transformers (tried using just HV, works in single player, fails in SMP) then back to the ground (70-80 blocks or so) to a HV transformer where it steps down from EV to HV and feeds a few sources from there. Right now it's a 3 way split to a MFSU, a MFE and the mass fab. In general just enough go to the MFSU and MFE to top them off and the rest goes to the mass fab.




    *used to be on the roof but i'm slowly tying everything together to one power grid

    Quote from FnordMan

    right now it's making ~75EU/t at the end reciving point. The source is 20 wind mills and 8 solar panels* on the top of the world.
    They go through a LV, MV and HV transformers (tried using just HV, works in single player, fails in SMP) then back to the ground (70-80 blocks or so) to a HV transformer where it steps down from EV to HV and feeds a few sources from there. Right now it's a 3 way split to a MFSU, a MFE and the mass fab. In general just enough go to the MFSU and MFE to top them off and the rest goes to the mass fab.




    *used to be on the roof but i'm slowly tying everything together to one power grid

    First thing is - you can't be losing 3-5 EU per 2048EU packet on 70-80 block long way. Exact packet loss for 3x Ins. HV cable is 56-64EU per packet when travelling such distance.
    You may as well consider using LV long distance cable setup, which can handle 128EU/t (your generators can make max 88EU/t) and loses up to 0%( :thumbup: ) EU. You won't need those 2 HVTs and MVT, so you can use them at more appropriate places.
    Setup looks like this: :Wind Mill::LV-Transformer::Cable::Cable::Cable::Cable::LV-Transformer::Cable::Cable::Cable::Cable::LV-Transformer::Cable::Cable::Cable::Cable::LV-Transformer::Cable::Cable::Cable::Cable::MFS-Unit:
    - not that expensive (1.2 copper per block of setup +some rubber and wood)(also you don't need diamonds)
    - you may argue that 84-96 copper bars is too much, but using diamonds, iron, redstone and other rare stuff is worse in my opinion. Everyone has loads of copper(not sure but i do).
    - loseless !

    10 geo-gens output 200eU/t total. Thus, sadly, you'd need two completely parallel copper lines run for this. However that isn't so bad; you still come out ahead on rubber even making the LVTFs.


    As for the EV situation, it does make sense if you need an extremely remote site for something, or if you didn't already know about the 128eU/t exploit for LVTFs (I'll gladly admit, I hadn't thought of it and presumed you needed to step the packets back up, thus making it even more annoying than it needed to be.

    Yeah, glad you mentioned that, I hadn't thought about the fact that 10 of them would overload the input capacity of a single LVTF and produce a backlog.


    Of course, that made me stop and think: Why upstep it from LV to MV to HV to EV?


    And I hooked up 13 geothermal generators to 4 pieces of copper cable (4 on each side, 4 on top, 1 on the end) feeding directly into an upstepped HV (skipping the upstep to MV and HV and going right to EV), running over 1 length of 4xHV, downstepping to HV, then feeding into an MFSU. Got 258048 out of 260000 in the MFSU. Exactly the same output as having those 13 geothermals feed into a pair of upstepped LVs, then an upstepped MV, then an upstepped HV, then back down to HV into the MFSU. With one difference, of course...the 13 geothermals backlogged the two LVs, so it was actually MORE efficient not to use them!


    Again, it's entirely possible I might be missing something here, but...why step up using 3 transformers when you can just upstep any voltage directly to EV and get exactly the same results?

    Quote from raGan

    First thing is - you can't be losing 3-5 EU per 2048EU packet on 70-80 block long way. Exact packet loss for 3x Ins. HV cable is 56-64EU per packet when travelling such distance.


    Wanna bet? input, as just measured a bit ago: input: 70 EU/t. energy output at the other end, around 65 EU/t. That's total loss after the conversion to EV and back to HV process. From there it gets distributed around my base.

    Measure it on the 4xHV cable connecting the transformers...what you're seeing is likely energy loss from the energy net on the receiving end, not involving the transformers at all. The transformers on your end will output full packets regardless of the energy loss on the HV cable...


    For example, I just set up in my test world the following system. Batbox - upstepped LVTF - 41 units of glass fiber cable - LVTF - batbox. Stuck a rechargable battery in it. Reading between first batbox and upstepper: 32 EU/t. Reading between transformers: 127 EU/t over 1 tick when I finally clicked fast enough to capture a packet. Reading between the receiving transformer and receiving batbox?


    128 EU/t over 1 tick.


    Energy was lost through the transmission system. You won't see that energy loss after the transformer, you have to read it between them.


    You're losing energy somewhere in addition to your transmission net.

    Basically, get yourself an EU reader, go between the transformers to the EV signal, and spam-click it until you capture a packet. You'll know you got it when you get a reading over 1 tick that was something other than 0. If your system was really only losing 5 EU/t over that distance, that packet should read 2043. I'm betting it'll read somewhere closer to the 1990 range.

    Quote from Nightdagger

    Again, it's entirely possible I might be missing something here, but...why step up using 3 transformers when you can just upstep any voltage directly to EV and get exactly the same results?

    Some say you can't use EV directly in SMP for some reason, but i have never tested it.

    Quote from FnordMan


    Wanna bet? input, as just measured a bit ago: input: 70 EU/t. energy output at the other end, around 65 EU/t. That's total loss after the conversion to EV and back to HV process. From there it gets distributed around my base.

    It's quite clear now. When sending 70EU/t, one EV packet is sent every 30ticks. You are not losing 3-5 EU per one packet, but per one tick. 65 out of 70 means you are losing about 7-8% of your input. 7-8% of 2048EU packet is 143-163EU. But as i stated above, only about 60EU per packet is lost during EV transmission, which is only about 3%, that means you must have greater loss somewhere else. Quite a big loss for that short distance.