4 ms·

In case anyone is interested, there are low cost (< $10k) portable diagnostic ultrasound machines. I used to work for a company that made interconnects for medi

by nhebb 10y ago

In case anyone is interested, there are low cost (< $10k) portable diagnostic ultrasound machines. I used to work for a company that made interconnects for medical devices, including ultrasound cables. Sonosite was developing their portable systems back then - 15 years ago - so low cost systems aren't something completely new. They don't have the same capabilities of full ultrasound systems, but they do fit the needs that some people are discussing here.

LeifCarrotson 10y ago
I have used a similar machine. It was purchased used for about $3,000 for a pregnancy resource center. And as a Test & Measurement EE experienced with some high-density analog cables, I was extremely impressed with the quality of the cable in that device. But reading your post and this one: > ... and then it's got to connect to the cables off to the system - all 100 to 200 wires down a cable that's a couple meters long, and thin/light enough for someone to use 8 hours a day without getting repetitive strain injuries. Do you know how thin that makes each wire? Try calling a few cable vendors and ask for a small diameter cable with 200+ connectors that will actually conduct a signal without significant crosstalk and let's see where you end up on price. It also mentions the system being mounted on a cart. This seems like an area where technology can make some serious improvements. Mount your ADC (for which speed, accuracy, power requirements, and channel counts have been steadily improving) within the probe, and stick a high-speed serialization/deserialization chip on the other end of it, and run power and digital data over a much smaller, cheaper cable. That's the direction that industrial I/O is going, at least. I guess the point of all this is that there are opportunities to make significant changes to the way these devices operate that will make them easier to build. Switch out the ASICs for GPUs, the analog cable for digital, the custom computer for commodity hardware, and so on because these technologies are more capable than they were in the past.
- temac 10y ago
  That advice is so good that people are already working on it... even in the past! Believe it or not, there is no plot for using outdated technology and keeping the prices high. This is a quite competitive market, and engineers working in the field are not stupid. There are already some machines using GPUs for the signal processing. I think there might also be some machines with ADC in the probe, although this is not simple (you also have to put the transmit side, and the rx side has to be of good quality) and you can really benefit from extremely high acquisition rates, so depending on what you want to do such a digital link is not trivial + plus you basically duplicate a large part of the machine in each probe if you do that, so I'm not even sure this will be better in every cases... Also; a new machine can take several years to design and bring on the market.
  - LiesNStartupPR 10y ago
    Exactly, thanks for putting this so succinctly.

by nhebb 10y ago

LeifCarrotson 10y ago

I have used a similar machine. It was purchased used for about $3,000 for a pregnancy resource center. And as a Test & Measurement EE experienced with some high-density analog cables, I was extremely impressed with the quality of the cable in that device. But reading your post and this one: > ... and then it's got to connect to the cables off to the system - all 100 to 200 wires down a cable that's a couple meters long, and thin/light enough for someone to use 8 hours a day without getting repetitive strain injuries. Do you know how thin that makes each wire? Try calling a few cable vendors and ask for a small diameter cable with 200+ connectors that will actually conduct a signal without significant crosstalk and let's see where you end up on price. It also mentions the system being mounted on a cart. This seems like an area where technology can make some serious improvements. Mount your ADC (for which speed, accuracy, power requirements, and channel counts have been steadily improving) within the probe, and stick a high-speed serialization/deserialization chip on the other end of it, and run power and digital data over a much smaller, cheaper cable. That's the direction that industrial I/O is going, at least. I guess the point of all this is that there are opportunities to make significant changes to the way these devices operate that will make them easier to build. Switch out the ASICs for GPUs, the analog cable for digital, the custom computer for commodity hardware, and so on because these technologies are more capable than they were in the past.

temac 10y ago
That advice is so good that people are already working on it... even in the past! Believe it or not, there is no plot for using outdated technology and keeping the prices high. This is a quite competitive market, and engineers working in the field are not stupid. There are already some machines using GPUs for the signal processing. I think there might also be some machines with ADC in the probe, although this is not simple (you also have to put the transmit side, and the rx side has to be of good quality) and you can really benefit from extremely high acquisition rates, so depending on what you want to do such a digital link is not trivial + plus you basically duplicate a large part of the machine in each probe if you do that, so I'm not even sure this will be better in every cases... Also; a new machine can take several years to design and bring on the market.
- LiesNStartupPR 10y ago
  Exactly, thanks for putting this so succinctly.