John Curl's Blowtorch preamplifier part II

Status
Not open for further replies.
Scott, not everybody understands the subtleties of low noise design. Gentle correction is the best course.
PS, I checked with Constellation and we are probably using 5ohms in the Expensive version rather than 10 ohms. It is been a year or so since we last talked about it, so nobody can remember for sure at the moment.
 
Waly, you got that part right, but remember this is a differential input so you get 1.25 ohms X 2 because the input devices are in series and get 2.5 ohms added input noise, still some significant contribution.
The input pairs for my BEST design are the 2SK146-J73, that I personally supplied Constellation and they are equivalent to 2 X 2SK147 for each individual device, so that gives me a 3dB advantage over the 2SK170-74 all else being equal.

Nope, this is diff in diff out and please note I said 10ohm/8 not 10ohm/16 (the total number of devices). It's 1.25ohm and if you really need 8devices per leg to get 0.4nV/rthz that's likely due to the zero PSRR. You need a power supply with insanely heroic low noise to prevent contaminating the gain stage, your usual source follower is far from optimal. And at least you could cascode those jfets to a few volts Vds, to avoid some excess noise at high Vds.

Bottom line, you used a way to large number of your jfet secret stash. You could do much better, but I know you stick with what works for you. Carry on, this is your fashion design, not sound engineering.
 
Scott, not everybody understands the subtleties of low noise design. Gentle correction is the best course.
PS, I checked with Constellation and we are probably using 5ohms in the Expensive version rather than 10 ohms. It is been a year or so since we last talked about it, so nobody can remember for sure at the moment.

Uh, "probably"?

se
 
diyAudio Member RIP
Joined 2005
Fighting as I am right now trying to cope with a product development team abysmally lacking in any notion of document control, I sympathize. I specified a C0G cap that is central to the product performance. When I did so there was stock. Later there was no stock (and not because they ordered the parts!). So I redesigned things to accommodate a part that seemed to be more available (and there was stock). Today I am given a BOM with highlights of long lead time parts and notice that the cap (not highlighted) is suddenly a 5% tolerance part!! And guess what? There is no longer stock on the correct footprint part!
 
bcarso,
I'm surprised you aren't supplying the BOM parts list with specific requirements. Why would you leave this to an uninformed team that probably is more interested in the industrial design or price point? I'm not saying you have to lock them into specific suppliers or exact part #'s but perhaps that is what it will take to achieve your design goals. Got to love the KISS principal but sometimes you have to take control and lock things down.
 
Member
Joined 2002
Paid Member
Fighting as I am right now trying to cope with a product development team abysmally lacking in any notion of document control, I sympathize. I specified a C0G cap that is central to the product performance. When I did so there was stock. Later there was no stock (and not because they ordered the parts!). So I redesigned things to accommodate a part that seemed to be more available (and there was stock). Today I am given a BOM with highlights of long lead time parts and notice that the cap (not highlighted) is suddenly a 5% tolerance part!! And guess what? There is no longer stock on the correct footprint part!

These days when you see parts you almost have to buy them all or at least enough for the future. Very happy to work where parts are considered better than money in the bank.
 
diyAudio Member RIP
Joined 2005
Steven the process is: my schematic with specific notations is used to generate another schematic and a BOM. But it seems that even when I proof the BOM it gets translated into another one with mistakes. There is also the tendency (or was until I raised hell) to look at Digikey, and if something doesn't come up immediately to substitute another part and not tell me (recall the story I told of discovering that X7R dielectric was substituted for C0G, which resulted in unacceptable distortion).

Anyway they are learning. Soon enough I cannot say yet.

Yes Wayne, I am now wishing that I had ordered the parts myself.
 
Member
Joined 2014
Paid Member
Steven the process is: my schematic with specific notations is used to generate another schematic and a BOM. .

Good to know that, even though CAD packages have had integrated CAM for decades and will spit out BOM, pick and place program etc companies can still add 3 layers of humans to screw it all up.

Also reminds me of reason #12 I got out of aerospace. When your FET supplier has the only 20 space qualified devices in the world for the next two builds and gives you 2 weeks to place the PO or they get thrown in the bin. As that Amplifier had marginal performance and the devices were eye-wateringly expensive this caused a lot of veeeeeery boring meetings.
 
billshurv,
Didn't you just love the Lean manufacturing thinking that said if you didn't use something for 6 months you tossed it out. In the tooling department I ran they did that before I got there and tossed tooling that wasn't even theirs. it was government property! My guys had to hide inventory they knew they would need so it wouldn't be trashed, it was very short sighted since there was unused room to store these things. Clean it out now and pay for special orders later at 10x's the original cost if you could even get something. I didn't have a problem with most Lean thinking but some of it was just simplistic shortsightedness.
 
Member
Joined 2014
Paid Member
well I was in the first wave of 'lean' (still have a copy of 'the machine that changed the world') and back then it was about being efficient, not chucking useful things. Sadly like all things it has morphed far away from what is was originally (japan rebuilding industry after a bit of a setback, helped by the fact that your suppliers were your family in many cases). A lot of good messages but all common sense to those at the coal face.

Mixing lean and life laundry tho... I have a maslin to make Jam in. I use it once a year so according to the life laundry lot I should throw it away each year and buy another one? I think not. And yes, I hoard. Not as bad as my Mum tho :)
 
billshurv,
My first week working in the facility and they threw me in as a co-leader of a Kaizen event for the platting department. I looked at the layout of the tank setup and figured out if they moved one platting tank they would triple the production rate. They told me that the TAKT time was much longer than they would be able to create the parts so they ignored the time savings, they didn't want to have someone standing around. Rather have the guy take more time and stay busy than increase productivity and have to find something else to keep him busy. Oh the stories we can all probably tell.
 
diyAudio Member RIP
Joined 2005
A good opamp, like the (now discontinued) LM4562 or similar makes an awesome power supply for a discrete front-end that does not have an intrinsically high PSRR. Do it right and you can expect >120 dB at LF and better than 90 dB at 100 kHz.
Meanwhile Hansen seems to believe he's the first to use voltage regulators that both sink and source current, according to a review in SP.

A friend was complaining about a VR part that was letting a lot of noise through from a USB +5V line, and he said it was much worse than another with seemingly the same PSRR. I looked at the datasheet and noted that the PSRR at 2kHz had a value of...wait for it...15 dB. I pointed out to him that yes, you have to look at graphs if they are provided.
 
Status
Not open for further replies.