Aion FX Anomaly (Hot Cake)

[Envee]

Try an OLD TL071.

TI has done a dirty deed. They released a totally new design under the moniker of a 45 year old design. It's not even the same kind of FET. They did-away with offset pins (which may be the secret trick on this pedal).

Thanks again.  I have a couple more questions to see if I follow you correctly. 
When you say "not even the same kind of FET," what is the change?  Do you mean that it used to be JFET only, and did not use BiFET technology?  I haven't yet found an archived datasheet that has unipolar output or anything besides bipolar output.  TI's datasheet is pretty unclear.  I don't see that it mentions JFET or bipolar output, but on sellers' websites it's usually described as JFET and sometimes spec details mention BiFET.  I found other manufacturers' datasheets from 1997, 2001, and 2005 that mention bipolar output. 
Sounds like you suggest I should look for even older chips, if I follow you correctly that is.  It does seem possible that Crowther has boxes of TL071 chips from the 80s and still uses those.  Might be tough.  I came up empty on Small Bear, Jameco, and elsewhere.

You also say they "did away with offset pins," so I was looking at the datasheet to find that.  As you say, the six-pin SOIC versions of TL071H omit the offset pins.  Also, the eight-pin SOIC TL071H version does not label pins 1 and 5 as offset. (See page 5 of TI's datasheet linked at the bottom, Figures 5-1, 5-2, and 5-3). 
But please see the eight-pin TL071 PDIP versions on page 6, Fig 5-4 (image pasted below).  That still labels pins 1 and 5 as offset, right? 


https://www.ti.com/lit/ds/symlink/tl071.pdf?HQS=dis-mous-null-mousermode-dsf-pf-null-wwe&ts=1685796025623&ref_url=https%253A%252F%252Fwww.ti.com%252Fgeneral%252Fdocs%252Fsuppproductinfo.tsp%253FdistId%253D26%2526gotoUrl%253Dhttps%253A%252F%252Fwww.ti.com%252Flit%252Fgpn%252Ftl071

[PRR]

The "BiFET" process was JFET input and maybe some small bits, Bipolar everything else.

ST LF351 2008 shows small value changes on the TL07x plan. It used to be more different (why TI stole the LF351's lunch).

The "new TL07x" is mostly MOSFET, little resemblance to the old TL0 chips.

[Envee]

Thanks a million!  Sounds like I need to find the date of the revision when the change to MOSFET occurred.  Shouldn't be as tough to find one before that date.

Also, holy cow.  These are advertised as JFETs, and I do not recall any mention of MOSFET.  I'll edit this post to add more pics of it being described as JFET.




https://www.ti.com/product/TL071?keyMatch=TL071&tisearch=search-everything&usecase=GPN-ALT#pps

[ElectricDruid]

Sounds like getting hold of an "old" TL071 is important for this circuit then. Reclaim them from scrap hi-fi from the 1980s.

(Oh great, another "mojo" part that people can obsess over! Just what we needed!)

[aion]

Try an OLD TL071.

TI has done a dirty deed. They released a totally new design under the moniker of a 45 year old design. It's not even the same kind of FET. They did-away with offset pins (which may be the secret trick on this pedal).

TI made a fairly nonsensical move by merging the H series in with the standard non-H datasheet, and in that process they replaced the existing diagram with the H series diagram outright even though it only represents a portion of the parts described in the datasheet. It would have been better to make a new datasheet for such a significant change (and even better, to use a whole different part number instead of a suffix). I expect it was management's bright idea for how to get the word out on the new series. But my understanding was that the non-H types are still manufactured the same as before, so if it's called a TL071CP then it should be the same BiFET topology as it's always been. I have a 2017 datasheet from before the H merger here for reference if anyone wants to check it.

From what I can tell, the H series doesn't even come in DIP-8, so we wouldn't be at risk of mistakenly using the wrong type for what we do. Hopefully we can avert a vintage chip crisis before it starts!!

[Envee]

Thanks very much!  That explains why PRR saw no offsets. 

I've continued to search for errors in my own builds.  I built a breadboard to see if that would help identify something.  Used some parts from different manufacturers (like cheaper caps).
But, I got virtually the same voltage results, with the same corresponding small voltage changes when the Mode switch is flipped.  Below are the voltages on my 2N3906/TL071CP build compared to the voltages on the breadboard [shown in brackets], with the slash to indicate the switch flip.  The breadboard also uses a 2N3906 and TL071CP (and same 2v7 diode). 

Pin 1: 0.14v / 0.14v  [0.14 / 0.14]
Pin 2: 4.75v / 4.67v  [5.1 / 4.98]
Pin 3: 4.32v / 4.24v  [4.63 / 4.52]
Pin 4: 0v / 0v  [0 / 0]
Pin 5: 0.14v / 0.14v  [0.14 / 0.14]
Pin 6: 4.75v / 4.67v  [5.10 / 5.00]
Pin 7: 8.56v / 8.56v  [8.25 / 8.25]
Pin 8: 0v / 0v  [0 / 0]

VC (Q1 Base): 7.63v / 8.56v  [7.5 / 8.26]
VB (Pin 3):  4.32v / 4.24v  [4.63 / 4.52]

So, can I ask what you all think?  Should I be trying to change something (R11?) that would result in a larger change to the VB bias?   Do the voltages above indicate that my builds and breadboard are not behaving like the schematic? 

Thanks again, all.  And I'd like to say again, the Anomaly is great.  The (non-Mad Bean) ones I've built match my Hot Cake on the settings that I use most often. 

[aion]

VC (Q1 Base): 7.63v / 8.56v  [7.5 / 8.26]
VB (Pin 3):  4.32v / 4.24v  [4.63 / 4.52]

Right off the bat I would say that voltage difference doesn't look like enough... when the mode switch shorts 2&3, VC should be 2.2V to 2.7V lower than VA since it passes through the zener, which would nominally drop 2.7V, but possibly less in this circuit due to the current. On both of yours VC is only about 0.8V lower than VA, almost like the zener is backwards (though it's slightly-enough higher than a standard silicon voltage drop that it's probably not actually backwards, it's just curious).

Which type of zener were you using? The lack of voltage difference looks like the source of the issue at least, but the cause might need a bit more digging.

[Envee]

Thanks very much for looking at this.  I saw Rob Strand's comment saying it may be 2.2v instead of 2.7v, but it's much less on my builds.  Just checked again - on all the builds VC is about 1v lower than VA and on the breadboard VC is about .8V lower.  So if I'm doing something wrong, I'm doing it consistently!
As for the orientation, the second post on this thread has pics of my 2N3906 build and an image from your website of the unpopulated PCB (cathode in square pad facing current, as in the schematic and breadboard). 

I'm using Mouser part no. BZX85C2V7-TAP, Vishay Zener 2.7 volt 1.3W 5%
https://www.mouser.com/ProductDetail/?qs=yIvl1B2wDi6nrjqtkvXm%2Fg%3D%3D
And, if I use the super-zoom on my phone camera, I see it says 2V7

I watched a tutorial and built a little circuit 9V+ --> 78r -->  cathode (2V7) anode --> ground, and it measures ~2.7V.  (80mA test current)
Did this test for the one in the breadboard and another from the package these came from.

The forward bias voltage drop is 0.72V. 
The reverse bias voltage drop is 1.475V.  I think I don't understand this.  When I checked the other zener, the 10V zener diode, the forward bias drop is .72 but the reverse shows "OL". 

[aion]

A lot of multimeters have a max reading of 2V for diodes, so the 1.4V reading would show up OK but the 10V would be overloaded (or over limit, or whatever else OL may stand for). Zeners behave as a normal silicon diode (0.7-0.8V) in one direction and the nominal voltage drop only comes into play when it's reverse biased, so those readings look like what I would expect.

I'm going to try to find that zener diode from the trace and confirm that it's a BZX85... the 10V one definitely was, but reviewing the trace now, the photo I took only shows "2V7" so it's possible that it's a BX79 or another type and I just assumed it was the same series. Looking at the datasheets now, the BZX85 series is designed for higher power applications, so it could be that another lower-power type like BZX79C2V7 (0.5W vs. 1.3W) might be closer to 2.7V in this particular circuit.

[Envee]

Wow thanks for getting so involved. I searched old orders but don't find that I ever ordered any other 2.7V zener diodes. All others are 10V and above. I'll check with some people around town that might have some stashed.
If you suspect another diode is a candidate, I'll add it to an order and put it in the circuit.
Thanks

[GGBB]

Yes, LF351 too. I remember the TL07x series being significantly simplified (cheapened) from the LF parts. Worked a lot alike but LF cost more. I thought at some point they quit the LF and put that number on the TL0 parts. The 2008 ST datasheet for LF351 sure looks like a TL071 schematic, but I forget details.

The current TL081 also looks like the OLD TL071: https://www.ti.com/lit/ds/symlink/tl082h.pdf.

[PRR]

I think aion has it in reply #24--- The "H" suffix is the "too new too improved" product.

And may not be sold as DIP (with legs).

But TI's datasheets are just getting worse. On the new TL08x sheet page 29, we see:

• 8.1 Overview
  "The TL08xH family (TL081H, TL082H, and TL084H) is the next-generation family of the industry standard TL08x
  (TL081, TL082, and TL084) high-voltage general purpose amplifiers.
• 8.2 Functional Block Diagram

The "block diagram" is a barely simplified schematic of the old non-H part.

Also page 25 Figure 6-44. Maximum Peak Output Voltage vs Load Resistance is still unlikely (output goes to dead-zero at 100 ohm load??). This error runs well back into the 1980s.

[ElectricDruid]

Pfft! No-one reads datasheets anyway, right?! 

[GGBB]

For the curious - every TI TL07xx data sheet to date since August 1994: https://ufile.io/dpiapi5t. (I was bored this morning.)

[Envee]

Ordered a batch of 1n5223B zener diodes, 1/2 watt

https://www.digikey.com/en/products/detail/vishay-general-semiconductor-diodes-division/1N5223B-TAP/4825636?so=81939369&content=productdetail_US&mkt_tok=MDI4LVNYSy01MDcAAAGMc6JBM3qrCL2zWL6d2uiEZgYvrF_bHLvlGtzKDfj7z9Ao19bRTNRxLGn6QXouAEL8TogFI2fE2xmgTaQhEQs1lJgg1th_jPhteKfC3nq1

[aion]

Just a general update - I grabbed another Hotcake of the same variant (internal slide switches) and measured the voltages against my prototype in both modes.

Note in advance that normal mode shorts the 2V7 zener. Bluesberry mode engages the 2V7 zener and puts a 220k resistor in parallel with 82k. I may have described this inconsistently earlier in the thread, but just for clarity going forward, that's what is happening in each mode.

ANOMALY
VA voltage: 8.98V (9.66V supply minus 0.21V for the series Schottky and 0.47V from the 220R series resistor)
Normal mode VC: 8.98V (same as VA)
Normal mode VB: 4.92V
Bluesberry mode VC: 7.53V with BZX79C2V7; 8.06V with BZX85C2V7
Bluesberry mode VB: 4.71V

HOTCAKE
VA voltage: 9.19V (9.66V supply minus 0.47V from the 220R series resistor, no Schottky)
Normal mode VC: 9.19V (same as VA)
Normal mode VB: 5.03V
Bluesberry mode VC: 7.66V
Bluesberry mode VB: 4.78V

So basically, taking into account the Schottky diode (which provides solid polarity protection at the cost of simulating a battery at 98% instead of 100%) all the voltages are close enough to be considered the same when using the BZX79C zener.

The BZX79C (0.5W) does in fact behave differently than the BZX85C (1.3W), dropping an extra 0.53V, and is much more in line with the voltages on an original unit. However, it doesn't drop anywhere near 2.7V on either unit - more like 1.5V. I'm not sure whether the 1N5223B would have the same performance since it's operating well below the specced amperage, but I would expect it's pretty similar since the wattage is the same.

In both my unit and the original, I could hear a difference in modes if I strained, but it was very, very subtle. I would echo what I said earlier in the thread about it likely being rig-dependent since some amps bring out differences in clipping more than others. But I'd be curious to know Envee's results after changing out the zener.