I've been working with the same method for a year without problems. Its mobile phase is:
A: ACN:H2O 33/67 25 mM NaH2PO4, 7 mM SDS, 1 mM ammonium acetate pH 6.2
B: MeOH
The gradient I use is:
min. 7= 60% A 40% B flow 0.7 ml/min
min. 9= 45% A 55% B flow 0.9 ml/min
min. 13= 60% A 40% B
I use a 15x 0.4 C18 Kromasil column with a c18 guard column.
A few weeks ago I started to have problems with the pressure. After injecting a few samples it started to down doing fluctuations. After purgering it dissapeared. At the beginning we thought it was a technical problem and we changed the inlet valve and the pistons but the problem continued although now it appeared after hours of work. After changing all the suspicious pieces of the pump we realised that if we work with the same mobile phase but without NaH2PO4+SDS+ammonium acetate the pressure is stable. After that we thought it is a buffer precipitation problems but I don't understand why it didn't happen before. The only change is the brand of the solvents but in all the cases they are HPLC grade and the mobile phase is filtered before being used.
I tried to use the same mobile phase in an isocratic way and the pressure is stable although the shape of the peaks are not very good.
I tried to include a cleaning step in the method with ACN:H2O:MeOH 20:40:40 (5 minutes) but I still have the problem after 4 or 5 runs (0ne or two hours).
Has anyone any idea what it is happening? I'm almost sure there is buffer precipitation (although when I mix the mobil phase in an beaker it doesn't precipitate) but why now?

If I understand what you write correctly (as in too many inquiries, some crucial info is missing, for instance, what is the analyte and its matrix?) the problem is obviously caused by what you inject.
As far as checking compatibility of mobile phases with each other and with sample one has to be very careful. I always pour one very slowly into the other in a test tube, watching especially the interface of the liquids, then mixing carefully while watching, adding more, etc.... Even if only a temporary precipitate appears one can expect problems on the column (frit). Problems of this type can change with slight, unnoticed differences in conditions.

Thank you for your answer but I don't think the problem is what I inject because the problem can appear without injecting anything.
My samples are plant extracts in methanol and I try to analyse alkaloids. Lately I've been injecting only standars (a mix of alkaloids in methanol)

i had qabout the same problem with a phosphate-buffer/methanol gradient. After lowering the concentration of phosphate in my buffer the problem disapeard. is this a option for you?
when you mix the mobile phase in a beaker then the mixing is compolete, but when your gradient mixer is mixing solvent it first takes a bit if solvent a then a bit of solvent B etc. so in your solvent line after your proportional valve there are "pieces" of pure A and pure B after eachother. in the place whre these pieces reach each other ther are parts with a lot of methanol and phosphate. That is whre the problem ocures. So if mixing in a beaker shows no problem, this thoes not allways mean no problem will ocure in you HPLC.

I tried to reduce the buffer concentration to 12.5 mM but I still have the problem so I don't know what to do. Any other idea?

It could be a problem, using a low-pressure gradient pump. Try to use a high-pressure gradient pump.

Do your pressure fluctuations coincide with baseline fluctuations? What is the magnitude of the pressure fluctuations? Try speeding up the flow. Do the pressure fluctuations pulse with the piston action?

Check the compressibility factors of each line. Make sure that they are correct based on an instrument audit.

Do you have any leaks?

Try adding an inline mixer between pump and autosampler.

What is your backpressure? Why do you change flow rate in the middle of your gradient? Try a uniform flow rate. Shock and recoil events are sure to follow from flow rate increases.

I'll wait for your answers before I ask more questions. There may be something to your choice of solvent brand names.

change piston seal