Dealing with Antminer S17 and T17 Issues

Hey everyone! So, I've been getting a lot of T17s and other miners sent my way. Figured it's time to kick off a new thread for all of us dealing with these machines. Sometimes I even get asked about fixing KNC Neptunes, but that's a whole different story. I'll share my experiences with troubleshooting these miners, tips to keep them running, and maybe a few rants about quality issues. With the ongoing chip shortages and the fact that even old models like S9s are still making profits, I think these miners are gonna stick around for a bit.

Reserved for future thoughts

So, first tip: don't ship these boards casually, okay? The design for S17 and S19 is a bit more fragile than the older S7/S9 models. The heat sinks are huge and only supported in the middle. If you press on them or if your shipping box gets a little rough, it could torque the chips and break them off. Not cool at all.

I gotta jump in here with a cleaning tip. Never use a strong air compressor on hot gear. I learned the hard way when I was dusting off an S17 pro; I blew the heatsinks right off. They flew off like nothing, and now I’m just waiting for things to fall apart completely. Would've preferred two more months of stable mining.

Right? The solder does soften up sometimes. Had a few boards where the issue wasn’t a bad chip, but just some extruded solder balls getting in the way. Did the heatsinks just pop off, or were the chips affected too?

For this gear, the chips are still intact, just lost the heatsinks. Not sure if the chips are damaged yet since it won’t run, but it logged all three hashboards fine. The kernel threw an over-temp error, so it seems the chips are okay, just can’t function without heatsinks.

I just removed some heatsinks using hot air at 360C without preheating. They came off pretty easily, and the solder layer is thin. You can reattach them at around 300C with a bit of flux.

Nice job! I just got a T17+ back up and running at about 45Th. I’ll share how it went down. Had a situation where two boards were cycling between zero and hashing, while the third was completely dead with only 33 dies showing. Diagnosed it, labeled the boards, and checked for dust and burns.

Good call, man! Cleaning is key. You gotta remove about 25-27 screws to really get in there. Dust gets everywhere under the controller case, and it can cause micro shorts in the PSU.

One more thing, about auto-tuning. Braains loves to auto-tune, but it doesn’t always do a great job. Boards can get shut down while it tries to mess with voltage and frequency. I prefer to turn that off and stick with fixed frequencies.

Yeah, I’ve had a similar issue. Even with every tuning option off, it still tries to shut down the boards. When I set them at 400MHz, they run way slower. It’s like a glitch fest. T17s usually have binned chips, so that’s why they’re cheaper.

Wondering if all boards are getting the same volts? Maybe you can set frequencies differently for each board based on performance? I have an S17 that can’t auto-tune because one board overheats.

Voltage drops? Definitely a sign of loose connections. They switched to bolts for a reason; I’ve seen too many burned boards on eBay. Lowering the voltage really helps with temps and power usage. I dropped mine to 16 volts and it’s much happier now.

I’ve got six S17 Pros hooked up to Braains. I’ll test the voltage change soon. Set freq 400 at 17 volts was doing 37.5Th at 1434W, now it’s 38.1Th at 1310W with a nice temp drop!

So how’d you fix that board? It was stuck at 33 chips, right? The chip count starts from the bottom left and goes up in a zigzag. Looks like chip 33 wasn’t at the edge but one in. Those hot spots matter.

If you mean those boards with a single Vcore, I’ve said this before. Using the PSU for all the regulation saves a lot of efficiency. It’s why modern miners are doing so well.

I was thinking dropping the voltages helps with efficiency too. But I wonder if they have extra regulators for fine-tuning chip voltage? Whatever they have seems tiny compared to a real power supply.

Yay! I’m helping save the planet by making miners more efficient! When I get my Nobel Prize, I’ll definitely mention this forum...

In the meantime, I’ve noticed that one board throws an intermittent connection error. Took out the board, examined the chip pads, and found some junk on them. After a reflow, the pads look much better.

Meantime I've been watching the logs on this T17 and noticed that board 1 will occasionally throw this error: Looks like the chip after 14 is having an intermittent connection. Great, I love intermittent connections. So took the board out, put it on the preheater, took off the heat sink and here's a few pics: On the preheater with the sink for chip 15 removed: Close up of the chip. The side closest to the intake fans had crud on the pads and of course solder balls from a bad factory flow. Bad factory! After a reflow: The pads no longer have solder balls as the solder is adhered to both the chip *and* the board. Preheat will do that... And now the board is back in and running with 44 chips. I'm letting it tune as all three board are now in good shape.