Actually, capacitors C1, C2 are 1210 in the layout shown, but if the trace is thin enough and the PCB fab can handle it, it should be fine with 0805 and 0603 also. I have used 12-mil traces between the pads of 0603 resistors on an ENIG board earlier, and the PCB house had no problems, and I was able to hand-solder the resistors without problems. If the solder-mask is good enough (i.e. doesn't lift or have bubbles/inclusions), it works fine.
Crosstalk isn't an issue since it's at right angles, but there may be a small amount of parasitic capacitance (~ 1 pF or less) between the component, dielectric solder-mask and the trace.
If the trace between the pads is a problem, then a zero-ohm 1206 jumper or routing it on the solder side are options.
Edit: Regarding coplanarity, it's not a problem for the Panasonic ECHU 1210 parts for C1, C2. They have a raised flange at the pads which essentially lifts the central section of the capacitor clear off the board. Panasonic probably did this to allow solvent penetration below the device, but whatever the reason, it seems to be good idea for capacitors.
Last edited:
OK, must be time to send the old tri-focaled eyeballs out for calibration.
But the production problems described by VivaVee and rsavas seem to still apply to 1206 and 1210 packages. To be honest, I haven't had enough experience with those SMD sizes to consider whether the same practices would apply to their layout.
I don't recall if a trace under the component was a factor but I once spent time staring at customer returns through a toolmaker's microscope to discover absence-of-solder under component contacts. The insidious part was that the assemblies passed production test but failed after a short service - presumably, because the clean, new, component made enough physical contact with the (solderless) pad to establish electrical continuity at the time of manufacture, but eventually separated from the PWB during service. And to REALLY complicate the case, sometimes the symptoms would "clear while testing" - sometimes (not always) to reappear after the assembly was returned to service. Based on this behavior, somebody convinced the Senior Manglers that the inept and over-paid design engineers had created a design that was randomly breaking into oscillation. After finding several instances of dry solder pads, I showed them to the production engineer who said he'd fix the problem - and I guess he did, though I never heard how.
And, as you noted, it's not a problem in cases where the component body is deliberately inset from the end contacts. That's why it's not an issue with SOT23 packages: going from memory, I believe the official JEDEC standards call for at least a thousandth or two of daylight under the body when it sits on its leads.
It's just as true that it won't be a problem in a hand-soldering operation, where the operator forms a solder fillet between the PWB and component contact. (Assuming you can make the lead-free stuff behave that way.) This would be a "normal" manual assembly process - not the unnatural acts that design engineers have coerced components to perform during development and prototyping.
So I guess the two sides of this question continue to glare at each other over the wall, as they have for a couple of decades now.
Dale
I call it a wire!!
You probably can use those SMT test points that are "C" shaped". Designed for scope probes etc.
Some SMT current shunts are of the same construction as well!!
As a suggestion, use SMT comps with folded end terminations to get around the coplanarity issue.
Would expect DFM issues to be the last of the concerns.
I can vouch that Contract MFG's like to add a premium for any DFM issue that they uncover, as part of a design review. So start adding those imaging fudicials & other DFM features as well!!
Yes, I was thinking of those current shunts in fact.I call it a wire!!
You probably can use those SMT test points that are "C" shaped". Designed for scope probes etc.
Some SMT current shunts are of the same construction as well!!
As a suggestion, use SMT comps with folded end terminations to get around the coplanarity issue.
Would expect DFM issues to be the last of the concerns.
I can vouch that Contract MFG's like to add a premium for any DFM issue that they uncover, as part of a design review. So start adding those imaging fudicials & other DFM features as well!!
The attached files are a documentation package for a proposed implementation of the SW-OPA Discrete Opamp. This documentation is a revised version of Post # 2620 in this thread, at < http://www.diyaudio.com/forums/anal...screte-opamp-open-design-262.html#post3341418 >.
The attached files incorporate revisions based on comments received in response to the previous post. Significant changes:
1. The input stage balance potentiometer has been relocated to the drain circuit of the input stage. Associated component values changed accordingly.
2. Protection diodes added to the output stage.
3. Minor component value changes based on reported parts availability.
4. Reference designators reassigned.
5. Thick-film resistors replaced by thin-film resistors in Parts List.
6. Ground plane removed from vicinity of input pins.
Changes #1 and #2 resulted in significant changes to printed wiring board layout and routing.
Dale
The attached files incorporate revisions based on comments received in response to the previous post. Significant changes:
1. The input stage balance potentiometer has been relocated to the drain circuit of the input stage. Associated component values changed accordingly.
2. Protection diodes added to the output stage.
3. Minor component value changes based on reported parts availability.
4. Reference designators reassigned.
5. Thick-film resistors replaced by thin-film resistors in Parts List.
6. Ground plane removed from vicinity of input pins.
Changes #1 and #2 resulted in significant changes to printed wiring board layout and routing.
Dale
Last edited:
"I agree with Mr. Marsh. My 62 year old eyes and fumble fingers can't handle such small parts. Ray "
If you guys want me solder up a few boards for you, then send me the parts & I'll do it for you. I can test them for operation as well, so you do not get DOA's.
This is great thread, so i can offer help to some good contributors.
At 53 and being very near-sighted, this is a cinch for me. Lots of experience doing 0.5 mm pitch, no problem. That and my 10x magnifiers, thin solder, fine tips etc.
Send me a PM and I'll supply my address, I live around Toronto,Canada
Cheers
Rick
If you guys want me solder up a few boards for you, then send me the parts & I'll do it for you. I can test them for operation as well, so you do not get DOA's.
This is great thread, so i can offer help to some good contributors.
At 53 and being very near-sighted, this is a cinch for me. Lots of experience doing 0.5 mm pitch, no problem. That and my 10x magnifiers, thin solder, fine tips etc.
Send me a PM and I'll supply my address, I live around Toronto,Canada
Cheers
Rick
Thanks for the offer! I will be laying out a thru hole board based on Scotts's two board sandwich idea for to92 devices and RN60D Vishay resistors. It will be 2 inches by 1 inch and have the Pacific Recorders 990 footprint with those gold pins. It's more in line with the FM Acoustics and the Jeff Rowland modules. And gee I wish it would work on +& - 24v which means output stage transistor substitutes like the 2n5551 and 2n5451 and mje171/181 but I'll get by with +&-18v. But I'm not an engineer and don't have the expertise to do that. I have almost all the parts as per SW, waiting on a few more. But I wish the part substitution test phase would get over (parts are multiplying rather than diminishing) and gee I wish someone would breadboard a pair and listen to the thing and maybe make comparisons to some of John Curl's circuits or the Jensen 990. There is an FET front end 990 now on the market. Ray
I believe this schematic is entirely compatible with the latest version. Please point out anything that appears incompatible.
As I mentioned in Post #2620, my implementation includes several design features intended to increase the experimental versatility of the assembly. For example:
- The DC level-shift devices (Q7, Q8, Q13, Q14) are implemented as transistors in SOT23 packages, even though the basic circuit function is technically a diode. If you prefer a true diode rather than a diode-strapped transistor, the PWB pads will accept BAL99 and BAL74 (depending on polarity) diodes.
- Maintaining each transistor as a single SOT23 device offers the greatest choice of alternate devices. (However, using multiple-transistor packages (e.g., BCM847/BCM857) would save PWB area, and improve thermal tracking.)
- The resistors in series with each compensation capacitor (R16, R10, R14) were eliminated in late design iterations. I retained them in the schematic to offer the greatest versatility in creating custom compensation schemes. To implement a simpler compensation scheme, bridge these locations with a small-value resistor (e.g., 1R0), or a zero-ohm jumper, or a scrap of found-on-bench (FOB) hookup wire.
- For that matter, the secondary compensation networks (R10/C2; R14/C3) don't appear in the latest iterations. There is no disadvantage to simply leaving these positions on the PWB unpopulated.
- I mentioned a few other examples in Post #2620.
I placed the input balance adjustment in the drain circuit of the input stage, in response to a comment following my post #2620. I don't think the details of this feature were posted in a schematic. If my implementation is still incorrect please show the correct implementation. (I moved it from the source circuit - the most significant change from the previous version.)
I don't recall any posted schematics that showed the output protection diodes (D2A, D2B) in place. The connections I show may not be correct.
I have adjusted the specified component values in a few places. I did this to increase parts commonality (fewer lines in the BOM), or to align the BOM with component availability at popular distributors. (Yes, this can change on a day-to-day basis.) As mentioned in some previous posts neither the exact resistor values, nor the transistor types, used in this architecture are especially critical.
I started putting together this documentation package in response to requests for a complete, consolidated, schematic and parts list suitable for DIY construction. I'd like it to be as up-to-date as possible, keeping in mind that there are several sets of conflicting requirements represented in this thread and you can't please everybody. If I missed something in nearly three-thousand posts, please point it out with a specific reference so I can efficiently incorporate it. If you can produce equal-or-better documentation, contact me via PM and we can discuss sharing the native P-CAD 2002 source files. Similar to Scott's comments when he semi-retired from the project, I have to point out that I have well over a man-month of effort invested in these packages and I'm quite willing to share the workload.
Dale
Ray, I agree that a mostly thru-hole implementation is in order.
I'd like to see Richard Marsh's comments on your idea before you invest a lot of time in it. He (and a few others here) want an implementation that can be easily probed, components removed and re-installed with minimum effort, etc. I don't know if packing some 50-odd components into a two-layer sandwich of those dimensions meet their requirements. I was thinking it would take at least 2"x2" on a single board; maybe as much as 2"x3".
(For example, the DIY version of a JE-990 shown in the photo at http://mnats.net/JE-990.html doesn't look like it's very accessible or modify-able - and the parts count is significantly less than SW-OPA.)
Would you lay the RN60's flat (more accessible), or stand them on end (less real estate)? Would you keep TO-92's in the output stage, or make provision for TO-126/TO-225 parts? (I think you can still get a few TO-92 parts with the "ECB" pinout that easily inserts into a TO-126 footprint.)
There's a Data Sheet for the PR&E 990 at http://www.technicalaudio.com/pdf/J...nd_related/Pacific_Recorders_JE-990_opamp.pdf . It shows pins for "SHIELD" and "FEEDBACK" that I don't fully understand. I expect the PR&E is essentially Deane Jensen's original 990 but I'd like to see a schematic to confirm that. Can you point me to outline drawings for these other modules?
Yeah, 24V rails would match the JE-990 specs. I think the SW-OPA input stage would also require modification to live with 24V rails. Perhaps not, if you limited the input common-mode range. There may be some posts about this a hundred pages or so back, but it's probably easier to investigate the concept in simulation than to find the posts.
Dale
That must be one of the Sonic Imagery Labs Enh99x Ticha series.
I have also finished the layout of a cascoded FET-input, SOT89 BJT-output FET990-derivative which is similar to the schematics I posted earlier in the thread with separate LTP and VAS CCSes. It differs in several minor details from Scott's and kgrlee's versions.
I managed to shoehorn it into DIP8, 15mm x 15mm, dual-sided SMD, O805 passives only. There is a single Rohm PNP dual for the current mirror, but everything else is bog-standard multiple-sourced parts. I could possibly replace the dual PNP with 2x SOT23 at a small cost in board area, will consider that on the next respin.
I've submitted the Gerbers for 30-day fabrication, and I'm waiting for them to revert and confirm. I can slip in a respun set of Gerbers during the hold if I'm able to complete it by then.
BTW, what's a good suggestion for an SOT23 PNP VAS transistor? I have MMBT4403, MMBT5401 and BC856 on the shortlist, but if there's anything more linear I'd be happy to substitute it. I have a few sample Toshiba 2sa1198-Ys with me, but the Vceo is marginal at 30V.
"...... I think the SW-OPA input stage would also require modification to live with 24V rails. Perhaps not, if you limited the input common-mode range...." Dale
The input stage is very much cascoded and would seem much more able to handle 48 volts across it but that is mere conjecture on my part and I wish some engineer savi soul would look at this seriously for higher voltage operation. I deal with spatial relationships well being an architect and pc board layout is up my alley.
So far I've laid down parts. I'm leaving off all compensation components as well as power supply decoupling devices and input filter or input anti-latch up diodes or catching diodes on the output to be outside the module on the mother circuit board. I am worried about shields on those input pins to the jfets, them being very wide bandwidth devices. All matched pair input to92 jfets and current mirrors are bent over and their flat sides epoxied together as well as the diamond output bias to output device for good thermal coupling. I'm going to use to-92 pads for the output devices. The to126 pad for the mje171/181 has the base on the outside rather than in the middle which I'm uncompfortable with. The mje devices I'm been told are much slower. The boards will be ground plane with thru hole plating. This will allow some hybrid SOT parts like those matched diodes to be used on the component side. I have a drawing of the Pacific Recorders 990 buried somewhere and when I find it I'll post it or I can email it. I don't have a pic or pdf URL account to cache pics. I want to buy few SOT parts as I have a attic full of thru holes parts I have been collecting for 30 years, and many matched 2n4401/03, 2n5551/2n5401 pairs and don't want to go thru the pain of matching again. I'm using Scott's D5BB final schematic. Thanks, Ray
The input stage is very much cascoded and would seem much more able to handle 48 volts across it but that is mere conjecture on my part and I wish some engineer savi soul would look at this seriously for higher voltage operation. I deal with spatial relationships well being an architect and pc board layout is up my alley.
So far I've laid down parts. I'm leaving off all compensation components as well as power supply decoupling devices and input filter or input anti-latch up diodes or catching diodes on the output to be outside the module on the mother circuit board. I am worried about shields on those input pins to the jfets, them being very wide bandwidth devices. All matched pair input to92 jfets and current mirrors are bent over and their flat sides epoxied together as well as the diamond output bias to output device for good thermal coupling. I'm going to use to-92 pads for the output devices. The to126 pad for the mje171/181 has the base on the outside rather than in the middle which I'm uncompfortable with. The mje devices I'm been told are much slower. The boards will be ground plane with thru hole plating. This will allow some hybrid SOT parts like those matched diodes to be used on the component side. I have a drawing of the Pacific Recorders 990 buried somewhere and when I find it I'll post it or I can email it. I don't have a pic or pdf URL account to cache pics. I want to buy few SOT parts as I have a attic full of thru holes parts I have been collecting for 30 years, and many matched 2n4401/03, 2n5551/2n5401 pairs and don't want to go thru the pain of matching again. I'm using Scott's D5BB final schematic. Thanks, Ray
I would be interested in a thru-hole pcb that fits inside a standard 19 inch rack chassis. Decoded, that means a large layout. I have no need to cram SW-OPA circuits into an iPOD case or other tiny space.
Mineature size is great for saving space/real-estate which translates into money savings. But that is to be done in high volume and automated production factory. Otherwise, there is no need for such small parts in a DIY project.... But if someone will sell me a completed tiny pcb SW-OPA then itsy-bitsy tiny size doesnt matter. Just trying to be practical and realistic.... because I know that I will never make it nor hear the SW-OPA any other way.... Not in this life time. Challenges are great but not this way with manually installing many dozens of smd's. Call me a wimp... a pussy. I just aint going to do it.
-Richard Marsh
Mineature size is great for saving space/real-estate which translates into money savings. But that is to be done in high volume and automated production factory. Otherwise, there is no need for such small parts in a DIY project.... But if someone will sell me a completed tiny pcb SW-OPA then itsy-bitsy tiny size doesnt matter. Just trying to be practical and realistic.... because I know that I will never make it nor hear the SW-OPA any other way.... Not in this life time. Challenges are great but not this way with manually installing many dozens of smd's. Call me a wimp... a pussy. I just aint going to do it.
-Richard Marsh
Last edited:
I can develop my own thru-hole pcb layout faster than I could wire up the circuit with smd.
I would be more interested, at first, in exploring the open/closed-loop controvercy and thus be cahnging parts and listening and testing. Not to replace 990 in some equipment I dont own.
Larger size will have No impact on audio. Layout might.
Do what makes you happy. That's what I do.
I would be more interested, at first, in exploring the open/closed-loop controvercy and thus be cahnging parts and listening and testing. Not to replace 990 in some equipment I dont own.
Larger size will have No impact on audio. Layout might.
Do what makes you happy. That's what I do.
Last edited:
Scot's latest front end, or input stage, schematic was posted on post #2643 and the output stage was posted on post #2593. Both are different from the schematic you posted.