I may have to redo several film speed, proper proof and dev time tests. The non-discolored batches seem to be quite a bit more active.
I'm with Thomas, that does seem crazy. I've used nothing but Patterson graduates forever and they were never ever the problem with anything. I can't see how they would be, unless they were not clean.
The other possibility..one that may be more likely is that I measure the stock solution with pipettes and a pipette pump. The two latest batches were both made with one of the pipettes being a new one. Perhaps the one it replaced was contaminated. I do my best to really clean them but they're hard to clean. Perhaps that had residue from my 510 Pyro days. The two pipettes are marked and consistently used for only the A or the B solution.
When I used Pyrocat I'd always make 1 liter or 2 liters of working solution, meaning 10ml or 20ml of concentrate A or B at any given time, which is super easy to do with a small graduate.
Never used pipettes, so I didn't encounter having to clean one either.
Hope you found your culprit.
I started mixing Pyrocat HD by acquiring the powder version packaged by the Photographers Formulary and distributed by The Imaging Warehouse, former Nova Darkroom. The Photographers Formulary version is just slightly different from the version Sandy King publicises at the PyrocatHD website (http://www.pyrocat-hd.com/html/mixing.html) because it uses 0.2 mg of Potassium Bromide for 100 ml solution while Sandy King uses half of that, i.e., 0.1 mg of Potassium Bromide.
The first time I mixed it I used demineralized water for both solutions A and B. Both came out pretty much colourless. Whenever I made a working solution it became just very slightly brown or sepia. Once I tried it using trays to make some densitometry tests and had it in the developer tray for almost 50 min. When I turned the lights on I noticed it had become slightly purple, almost lavender. I attributed it to oxidisation. After about 2 months, the small flask for solution A also started to turn lavender. I decided not to use it and instead bought some glycol to mix fresh solutions for A and B, with just the A mixed in glycol. With glycol I have not seen solution A turn lavender. All of these has made me pretty much convinced that the color change was all about oxidisation. The more so because the sepia tone is stronger with solution A mixed in glycol.
I must say that mixing PyrocatHD is both lengthy and laborious. I use an IKEA cappuccino blender to mix it and it takes several minutes to mix, specially solution B because it requires slow addition of the Potassium Carbonate. A long mixing session of solution A can lead to extended oxidisation.
I really don't think the issues reported could be due to a specific tank, unless, as others have stated, it is not properly washed.
If anyone has any idea of the differences between the Photographers Formulary version and Sandy King's version and if these are important please "chip in".
While I have never experienced the shift to a lavender color I can agree that may well be an early sign of exhaustion, hard to believe the vessel is at fault, rather the vessel maybe somehow contaminated
I mix the Sol. A and store in a 1000ml bottle and store for most times 4-6 months, never have I seen a shift away from clear solution.
When the bottle does get around 1/3 full and I have a large amount of film to D from an extended trip I would mix another 1 liter and combine the two and work from that same stock solution knowing that each A & B solutions would extend through the D of the entire run of film.
I still use the original PyroCat HD, have tried the two bath scenario with low density results but do plan on tweaking that technique through trials to take advantage of shorter D times.
This is premixed from Formulary, in glycol. And the batches that haven't turned purple suggest it's not exhausted. We'll see how things proceed.
I use a separate syringe (with plastic tip) to measure both A & B and rinse them thoroughly with hot water immediately after use. They are cheap, easy to clean and seem to be quite accurate.