I don't know what your application is, but might glutathion be of any use to you? It is a tripeptide containing an -SH group and readily forms dimers via disulfide bridge.

What is the acceptable molecular weight range of the dimer?  Are you trying to isolate or identify this dimer in a mixture or you want a known dimer that you can purchase or generate?

One step up in complexity from glutathion would be a disulfide-stabilized leucine zipper http://www.ncbi.nlm.nih.gov/pubmed/9376362, while disulfide stabilized antibody Fv fragments (dsFv) have the added complexity that each domain additionally contains an internal disulfide bond https://books.google.ch/books?id=GDuy5zL6eRAC&pg=PA153&lpg=PA153&dq=disulfide+stabilized+leucine+zipper&source=bl&ots=bgnFmxRzGf&sig=li_2e5L_1NJn8qtyA0iJLZXnL3c&hl=de&sa=X&ved=0ahUKEwjX1rey0tbMAhVCbiYKHeAyAEAQ6AEINjAC#v=onepage&q=disulfide%20stabilized%20leucine%20zipper&f=false

Thanks for the fast answers!

Glutathion is indeed to simple. We are working on a magnetic TCEP analogon and allready have successfully tested (i.e. reduced) Ellmans reagent. At the moment we are reducing Insuline but instead of 2 peaks of the reduced dimers we get 5 or more. Not too nice.

So ideally we are looking for something below 50kDa, commercially available..

Is the leucince zipper commercially available? Another possibility would be FABs but they are expensive..

It seemed very likely that this would be your situation but there are possibilities.  If you don't mind a few intra-disulfide bridges, the TGF beta superfamily may offer numerous possibilities but it depends upon budgeting and the amount needed.  This group of proteins are cysteine knot covalent homodimers ~30kD-40 kD having one inter-disulfide bridge.  So this is a bit better than insulin with 2 inter-bridges and the one intra-bridge in the A chain.  There are numerous members of each sub-family:  TGF beta's, BMP's, GDF's, GDNF's, Activins, Inhibins, Nodal, Lefty, MIS.  If you're able to, you can drastically reduce the cost by your own production which I was able to do for GDF 15 from E. coli inclusion bodies as well as for other super family members at a former company.  Unfortunately I can't offer any free samples at this time but that doesn't mean this is not a possibility in the future.  

Another possibility is to purchase a relatively low-cost but pure monomeric protein that has just one inter-disulfide bridge and also has an internal (endoproteinase) clip site(s) which is located between the two cysteines, somewhat like insulin.  Treatment with the protease would generate the kind of simple dimer, albeit hetero,  you want.  This is often times encountered in recombinant protein purification and is a nuisance but needs to be tested and consequently may be a potential market for your product because of the ease of removing the reducing agent.  

By the way, after the reduction by the solid phase phosphine analogue is there an alkylation step or an acidic pH drop (protonate the cysteines) to prevent/slow re-oxidation of the insulin?  The reduction may have been successful but refolding post-reduction could have occurred to generate the 5 peaks.  Hopefully the reduction was not partial.