• Updated 2023-07-12: Hello, Guest! Welcome back, and be sure to check out this follow-up post about our outage a week or so ago.

Slow SPI (Half Speed) BlueSCSI DB25

68kPlus

Well-known member
Hi everyone,
I received my BlueSCSI recently from Bruce at RecapAMac in Australia, and I found that no matter what SD card I use, it always seems to run in half SPI speed (25MHz). While the speed difference isn't noticeable according to the creator of the BlueSCSI, it seems like my Mac Plus runs noticeably slow even with SCSI Accelerator 7.0, with only around 240KB/s Read and 210KB/s write.
Can anyone help?

Thanks!
 

cheesestraws

Well-known member
The only virtue of the BlueSCSI is its price. If you're expecting it to run fast or last a long time, you'll be disappointed.
 

cheesestraws

Well-known member
That's okay. How long do BlueSCSIs last?

Difficult to say, don't think they've been around long enough to tell yet.

The main problem here is that they drive the microcontroller far too hard: the amount of current that even a farily moderate SCSI bus expects to be able to sink through the device is about five times as much as the "absolute maximum" rating for the microcontroller. And the "absolute maximum" is "damage may occur instantly if you exceed this", rather than "maximum in normal use".

So, I mean, it'll work until it doesn't :). Which is fine so long as that's what you're going in expecting.
 

68kPlus

Well-known member
Difficult to say, don't think they've been around long enough to tell yet.

The main problem here is that they drive the microcontroller far too hard: the amount of current that even a farily moderate SCSI bus expects to be able to sink through the device is about five times as much as the "absolute maximum" rating for the microcontroller. And the "absolute maximum" is "damage may occur instantly if you exceed this", rather than "maximum in normal use".

So, I mean, it'll work until it doesn't :). Which is fine so long as that's what you're going in expecting.
That's interesting. Does it make a difference to have external power running into the BlueSCSI?
 

Fizzbinn

Well-known member
That's okay. How long do BlueSCSIs last?
The only virtue of the BlueSCSI is its price. If you're expecting it to run fast or last a long time, you'll be disappointed.

I have several BlueSCSIs, 2 desktop, 2 PowerBook and an external DB25 that I’ve had for between a year to 6 months, all still working. I don’t have a Mac Plus but have been very happy with their performance in my systems, seems on par with the SCSI2SDs I also have. I certainly don’t run them anywhere near 24x7 but I don‘t imagine many folks would in this hobby. Ultimately I suppose time will tell.

Of note there is also a F4 (faster processor/performance) Blue SCSI version that uses bus transceivers in order to not stress the microcontroller and as I understand it increase compatibility with non-Mac SCSI systems: https://androda.work/shop/

While not for everyone, I’ve also certainly enjoyed assembling my BlueSCSIs from less expensive kits. I also continue to be impressed with the firmware development performance and feature wise. In my experience the developers are responsive to troubleshooting questions and feedback too.
 
Last edited:

cheesestraws

Well-known member
Oh, good, so not all the BlueSCSI people are knowingly selling defective hardware, just most of them. That's... an improvement? Perhaps?

I'm afraid I lost patience with the lot of them some time ago, the combination of self-righteousness, anger and incompetence just isn't for me. Others can make their own buying decisions, of course.
 

ymk

Well-known member
Oh, good, so not all the BlueSCSI people are knowingly selling defective hardware, just most of them.

What concerns me, because I've been asked about running several devices on one bus, are the fixed resistor nets.

Simply floating common does not make the networks disappear from the bus!

Yes, sockets cost a bit, but hard drives use them for a reason.

As for OP's issue, a 25 MHz SPI link can move nearly 3MB/s, so I'd guess that's not your bottleneck.
 

erichelgeson

Well-known member
I'm not sure why every thread here about BlueSCSI has to have the same unrelated conversation when people are trying to get their macs going - the items brought up are never the issue people are looking for help with. You'll find friendly, helpful advice over on our github page or on tinkerdifferent.com.

Thanks to those that sent me this odd conversation. We'll get you sorted out via the other channels.
 

pfuentes69

Well-known member
It's really disturbing... I keep reading always these reactions when the word BlueSCSI is mentioned in this forum...

I'm still waiting to know about a BlueSCSI user that had one failing because of its design. What is clear is that is a low-cost device, which uses components such as the BluePill that aren't bullet-proof, but it's still the best ratio cost/functionality or cost/performance you can get!
 

cheesestraws

Well-known member
"It hasn't broken yet" is a pretty low bar to hold something to, and isn't really an answer to an engineering critique based on the structure of the thing.

People can buy what they want, but for me personally, it's the kind of bad engineering I'd be ashamed to be associated with. "It hasn't broken yet" doesn't change that, and is, from an engineering perspective, an almost completely empty statement.
 

ymk

Well-known member
the items brought up are never the issue people are looking for help with.

You don't see all of the support issues.

I've already brought this to your attention on TD and you've dismissed it, so my post was for others here.

Your board passes current between data lines when not addressed. For that reason, it shouldn't be used mid-chain.

Part of designing a SCSI device is making sure it doesn't interfere with other devices.
 

68kPlus

Well-known member
Hi everyone,
I received my BlueSCSI recently from Bruce at RecapAMac in Australia, and I found that no matter what SD card I use, it always seems to run in half SPI speed (25MHz). While the speed difference isn't noticeable according to the creator of the BlueSCSI, it seems like my Mac Plus runs noticeably slow even with SCSI Accelerator 7.0, with only around 240KB/s Read and 210KB/s write.
Can anyone help?

Thanks!
*UPDATE*
The BlueSCSI performs much quicker with another program, HDA Benchtest.
My results:
Sustained Read: 407KB/s - Low 397KB/s, Peak 416KB/s
Sustained Write: 220KB/s - Low 216KB/s, Peak 224KB/s

Seems like it was an issue with the program I used.
 

AwkwardPotato

Well-known member
I'm not sure why every thread here about BlueSCSI has to have the same unrelated conversation when people are trying to get their macs going - the items brought up are never the issue people are looking for help with. You'll find friendly, helpful advice over on our github page or on tinkerdifferent.com.
I'm not sure why you're insinuating that the advice over here is unreliable, seeing as the negative conversation regarding your product consistently comes only from the same couple of voices. That said, I'd imagine those voices would be less negative towards BlueSCSI were it not for your record of asshattery here and elsewhere.
 

Corgi

Well-known member
You don't see all of the support issues.

I've already brought this to your attention on TD and you've dismissed it, so my post was for others here.

Your board passes current between data lines when not addressed. For that reason, it shouldn't be used mid-chain.

Part of designing a SCSI device is making sure it doesn't interfere with other devices.
Is this a flaw endemic to the design? I already have three F4Lites on the way and I'm not sure if I want to put the one I bought for my Indy in the Indy if this is the case, because I don't think it's going to work and that hardware is significantly more brittle (and harder to repair) than Apple kit…
 

cheesestraws

Well-known member
Ah, I see I have been blessed once again with an ad hominem from the great St. Eric Helgeson, the Only Person Who Understands SCSI On The Internet, YouTube Approved™.

Let's be clear here . Here's a quotation from the draft of the ANSI/ISO SCSI-3 parallel interface standard for single-ended operation. (I don't have access to the final standard and I'm not buying it for the sake of this argument.)

Screenshot 2022-10-05 at 11.37.17.png
Let's look at the datasheet for a totally random STM32 part, under "Absolute Maximum Ratings":

Screenshot 2022-10-05 at 11.48.14.png

Blue Pills are so variable that obviously there isn't a single datasheet for it, so this will do. Let's take the 48 mA per signal in the standard. Let's be nice to the BlueSCSI and pretend there are only 8 data lines, and let's also be charitable to BlueSCSI and say that the actual current is 24 mA. Even this wild standards- and reality-defying underestimate is 192 mA, well above the absolute maximum rating of this chip.

And don't forget, when they say absolute maximum, they mean it. That is the amount of current that can pass through the chip before it is actively damaged.

BlueSCSI works by luck and by guess, is strongly susceptible to variations in the Blue Pill boards used, and is nowhere near standards compliant.

@erichelgeson - we have all cocked up designs. People here haven't seen my cockups but I have made them, dramatically and awfully. I have done really stupid things in my professional career, much worse than this. Mature engineering practice would be to fix it and move on. It's not a huge change. You could fix this in an afternoon. Instead of which you spend your life finding people on the Internet who give your product bad reviews and abusing them, which takes far more energy and achieves very, very little. Get over your insecurities and learn to be an engineer instead of a cult leader. Electricity doesn't care if you bluster. Only the concrete, measurable reality matters. Learn this and your hardware and your demeanour will both improve.
 

3lectr1cPPC

Well-known member
Is this a flaw endemic to the design? I already have three F4Lites on the way and I'm not sure if I want to put the one I bought for my Indy in the Indy if this is the case, because I don't think it's going to work and that hardware is significantly more brittle (and harder to repair) than Apple kit…
Of note there is also a F4 (faster processor/performance) Blue SCSI version that uses bus transceivers in order to not stress the microcontroller and as I understand it increase compatibility with non-Mac SCSI systems: https://androda.work/shop/
If this is correct than those shouldn’t have this design flaw. I bought an F4Lite for my PowerBook 180c and it’s worked great so far, I hope the F4s don’t because they’re just so much cheaper than any other SCSI device for PowerBooks.
 

androda

Well-known member
... I hope the F4s don’t because they’re just so much cheaper than any other SCSI device for PowerBooks.
F4 and F4Lite BlueSCSI use an active termination system which shouldn't cause cross-talk when used mid chain.

Some models have resistor packs that are removed to disable termination, and others use bus switch chips to move the terminator resistors out of circuit. Removing the resistors packs physically disconnects the signals so there can't be any crosstalk, and the bus switches perform effectively the same function with a solid state mosfet switch.
 

erichelgeson

Well-known member
Seems like it was an issue with the program I used.
I used the Silver lining speed test tool and got similar results to what you just posted, so back to the actual topic of the thread the SPI speed doesn't seem to be an issue.

I've already brought this to your attention on TD and you've dismissed it, so my post was for others here.
I didn't dismiss it, I agreed with it. It's the exact same design as RaSCSI. While old HDD did have removable resistor nets, it's not a great user experience. MacSD has "pull up" and "pull down" modes, which I cant find any reference to in a spec (seems that would only do 1/2 of what termination is described as doing). FPT is also not in the spec and wouldnt that overdive the MCU as it's done internally? I donno - I'm not an EE.

You could fix this in an afternoon. Instead of which you spend your life finding people on the Internet who give your product bad reviews
Honestly how would you fix this in an afternoon? It's opensource, you could actually fix it and open a PR! While out of spec there's been no BluePills being damaged in the way you describe they could be. Again I'm not an EE so I'll have to rely on the community contributing fixes or designs to this, of which you seem passionate about.

I'd imagine those voices would be less negative towards BlueSCSI were it not for your record of asshattery here and elsewhere.
I'm generally a nice person and generally people tell me that. Its only here people attack me, then I'm the ass for replying. Just trying to make the best SCSI experience for people with old macs as I can.
 
Top