TMC2100 vs. TMC2660

[Macport]

Hello,

My engineer/electrician is unavailable at the moment. I am attempting to understand these two chips. I have attached the datasheets for both and I have been wading through them myself.

- the TMC2100 is on the SilentStepStick and would be used in conjuction with the Duet 0.8.5 and RADDS

- the TMC2660 in on the new Duet Wifi

Both chips sport company jargon labels like microPlyer™, stealthChop™ and spreadCycle™ for the TMC2100 and stallGuard2™, coolStep™, microPlyer™ and spreadCycle™ for the TMC2660

For this scenario the two configurations would be implemented on a medium sized desktop printer and cost is not a factor.

My questions are, given this scenario:

1) Would there be a significant difference in performance?
2) Is their architecture and feature set different enough that one stands out over the other?
3) Has anyone ever had direct comparison of the performance of these two setups.

For these questions my focus would mainly be on printer performance as defined by speed, accuracy and resolution. Secondarily would be noise, WiFi vs Ethernet, etc.

Thanx,

Mac

[Eric]

There's a lot more to the equation than the driver chip. Even if the datasheet shows the chip supports a particular feature, that doesn't mean the firmware and/or hardware solution makes use of it.

[Xenocrates]

So, the biggest differences are in amperage capacity (The 2660 wins in theory, but it's doubtful that it could use it's whole 4A capacity in the Duet wifi. I believe they tested it to 2. something for loading and it worked), and in voltage tolerance, as the 2100 can go up to 46V, while the 2660 stops at 30. The 2660 also has intelligent current control to reduce idle temps, which is good. In short, for most applications which are not specifically engineered to stress them, or which are beyond the needs of almost any printer on the market, they are effectively identical. The 2660 can do heavier loads, the 2100 can go faster (More voltage means less rise time to where you're going), but you really won't notice a whole lot of difference between the two in any printer that's on the market at a consumer price.

For the boards you mentioned, the Duet .8.5 won't make much use of some of the features, such as higher microstepping, as the firmware for it assumes a locked x16 microstep for the most part (as that's what's on board), but the Duet wifi being designed with all that in mind, has firmware support for most all of the cool tricks.

If I were you, I would go with the Duet wifi solution. It will be faster (Processor wise), and easier to work with from a Human-Machine and Machine-Machine interface standpoint (removable SD card, and wifi support built in), and is a more modern design, which means it will be supportable longer, and comes with some other real noticeable QoL improvements.

[Macport]

Thank you Eric and Xenocrates for your quick responses.

My engineer/electrician friend will smile when he returns and reads this (smile as in laugh at me as he says "duh")(I get that a lot as the design/software guy). My intent was to ask for help in his absense and I have received excellent answers.

So the next question would be . . . is there any other 32bit configuration that can utilize the full potency of the TMC2100 that can match the Duet Wifi? (udoo, BAM&DICE, I don't know help me)

[Xenocrates]

As far as I know, the only other hardware that looks like it would do better (that's in production) is the Replicape with BBB, but Redeem has AFAIK, rather rudimentary delta support in comparison to the other 32 bit boards, and is missing some of the features I really like in the Duet, such as the PT100 support, as well as requiring two boards to start with (The replicape and beagle bone black), and potentially more, if you add the (as yet unproduced) new version of the Reach for support for more extruders, even before you add on a display (It does use a DVI over HDMI interface to connect to the display they like to use though). There aren't a lot of delta printers using it, and it lacks the auto-calibration that both of the common 32 bit boards have (Smoothieboard and the Duet, the Duet is better, but Smoothieboard had it first, especially for delta's, thanks to 626pilot (A regular of these forums)).

there is also the Smoothieboard V2 Pro, which hasn't had an updated ETA since the end of 2015 (As Arthur Wolf sometimes comes through these forums, he might have an update for us, but I also haven't been tracking it heavily), which uses the TMC262 driver, which is a higher voltage yet driver than the 2100, which means more potential speed, and nearly as high an amperage as the 2660.

[Macport]

Xenocrates - Again I thank you. I've been reading multiple forums for weeks if not months.

My intent is to flesh (suss) out what's best - in all terms.

I confess I've studied Replicate and Smoothieboard, but I wanted a response.

It's firmware and hardware board match.

Someone tell me this isn't a board to challenge industrial boards.

[Macport]

Okay Duet WiFi ordered.

Xenocrates synopsis of the viable boards (and their firmware features), within the parameters I described, reflects what my partner and I had found. I want to build this next machine over this coming month or two and Smoothieboard V2 isn't here to compare.

To go with that our next build has a 600 x 600 x 1200 mm build volume and we'll want what the Duet WiFi can bring to that.

All that said, it still comes down to the board/firmware match. The machine just finished owed a lot to Marlin's thinkyhead. This one will owe a lot to dc42 (If you just tally up the number of David's posts on all the different forums, the time commitment to tweaking the firmware, etc.).

Thanx all. I'll need to return here often to keep learning. I'll post if I think we've learned anything that can help the group. Photos and documentation in the fall when we start the big build.

[dc42]

The TMC2100 and TMC2660 both support microstep interpolation, for example you can send step pulses at the 1/16 rate and the chip will interpolate to 1/256. This is what Trinamic calls 'microplyer'. So both will make a printer much quieter, especially during long straight moves for which the interpolation works best. They also both support spread cycle and stealthchop modes. But in other respects they are rather different:

- The 2660 provides an SPI interface to the processor, as well as the STEP/DIR interface. This can be used for many purposes including programming the decay mode, setting the motor current, changing the microstepping, and monitoring the driver temperature. I plan to use most of these features in RRF.

- The TMC2100 is rated at 1.2A RMS current, 1.7A peak recommended current (less than many other popular drivers). The TMC2660 is rated at 1.6 to 2.2A RMS depending on temperature, so around 50% more. To achieve this, standstill current reduction is required, which we haven't yet implemented in RepRapFirmware. That is why we limit the peak current to 2.0A for the time being. That's still more than most other 3D printer controller boards can handle.

In summary, the TMC2660 is much more programmable, and it handles higher motor current. The TMC2100 handles higher voltage, but as 3D printers are invariably run from either 12V or 24V, and as the larger motors that might benefit from higher voltage almost always need higher current than 1.7A, I don't think that's useful in the 3D printer market.

The TMC262 that has been mentioned is a TMC2660 without internal mosfets. So a board based on it needs 4 external mosfets per driver, making it larger and more expensive than a board based on the TMC2660; but such a board will be able to driver really large motors. So probably more suited to CNC machines than 3D printers.