Stock Control

[thermionic]

Hi,

As some of you here run small OEMs, what do you use to keep on top of stock control?

I'm thinking that a tablet PC, running Excel, kept dedicated to the task might work? Every time someone removes or replaces an item of stock they update the tablet (yes - I know this requires discipline...).

In an ideal world, I'd be able to link it with Mouser / Farnell etc and get it to fire off a BOM.

What do you employ for the task?

Thanks in advance.

 

[ruffrecords]

I keep all my components in labelled drawers or boxes. When I use them, if the drawer/box is nearly empty, I write a note on my ToDo pad. In the evening, when I go through my emails, I check this list and order what is needed. OK for a none man band like me but not practical if you have staff unless you have a reorder list next to where the components are that anyone can add to.

Cheers

Ian

 

[thermionic]

I keep all my components in labelled drawers or boxes. When I use them, if the drawer/box is nearly empty, I write a note on my ToDo pad. In the evening, when I go through my emails, I check this list and order what is needed. OK for a none man band like me but not practical if you have staff unless you have a reorder list next to where the components are that anyone can add to.

Cheers

Ian

Thanks. That's what I do at the moment. The thing is, some components don't fit in drawers, so go in cupboards. I find myself ordering 'when I get low', which is a risky strategy, as some parts can take weeks to arrive. A spreadsheet that shows exaclty how many complete units can be built would help to prepare in advance.

 

[Rochey]

I'm going through this process at the moment, and will be showing my process on Expat Audio's facebook page. (Thursday nights, 9pm Central USA, live stream).

So far, I'm using a combination of a cloud based spreadsheet and a Zebra 2824 label printer. There is an interesting app called "Appsheet" that makes your spreadsheet into an app that runs on your phone. That app also integrates a barcode scanner.
Also, barcode scanners are now $20 at Amazon, they emulate a keyboard.

I'm also learning to write Python to use a barcode to trace a pcb through the manufacturing flow, from assembly, optical test, electrical test, programming, packaging and shipping.

Watching as many "lean manufacturing" videos on Youtube as possible, especially those by "Fastcap" are amazingly inspirational. Also, all of TI's EVM's that come from our contract manufacturer "Krypton" all come with tiny QR codes. That definately inspires what's possible.

Cheers

R

 

[Matt Nolan]

Only marginally related, but I just started working on a collaborative project and we are keeping track of the to-do list using a free App called Trello. Seems to work quite well, though it doesn't really work properly offline. Multiple users can add cards to lists, move them from to-do to doing to done, etc. and everybody gets updated. There are other apps which do the same or similar things. Could maybe be adapted to stock control.

 

[JohnRoberts]

While I don't do cloud, I wrote my own inventory control software back in the '70s (Basic).

Some old school inventory control is to use paper cards in each inventory bin, where you deduct quantity you pull. The card will show how much should be in stock and you can visually notice discrepancies, a reorder point or quantity to reorder can be on the card.  Of course this is old school manual control system.

Even I used my computer instead of paper, but this was before spread sheets and modern software. One inventory strategy that I used, was to perform a physical count when restocking a new shipment of parts. Instead of performing a time consuming physical inventory on everything once or twice a year, I count the parts in the bin when they are at their lowest level, just before adding the new parts.

I used BOM for the different products to let the computer pull groups of parts from raw inventory simply since I had numerous parts common to different products.

Trust but verify... trust the computer to manage inventory, but confirm that you really have the parts you think you do, when restocking for lowest impact on manpower.

Rather than counting large batches of loose components, with a precision scale you can weight say 10 parts, then weight the whole batch and calculate how many there are with simple math. This will generally be accurate enough and modern precise scales are inexpensive these days.

  JR

 

[ruffrecords]

I'm also learning to write Python

Cheers

R

Me too. Great language.

Cheers

Ian

 

[ruffrecords]

Thanks. That's what I do at the moment. The thing is, some components don't fit in drawers, so go in cupboards. I find myself ordering 'when I get low', which is a risky strategy, as some parts can take weeks to arrive. A spreadsheet that shows exaclty how many complete units can be built would help to prepare in advance.

Maybe you just need to keep a list of long lead items and check them regularly, whether you use them or not?

Cheers

Ian

 

[Rocinante]

Fwiw myself and a few kitchen managers run a multimillion dollar restaurant food, liquor, merch, and other stock on excell.
Used to strictly use paper now its about half tablet. There's several free apps that are easy to use. And trust me if this jack ass can do it, you should be able to with your eyes closed.

 

[mattamatta]

I find myself ordering 'when I get low', which is a risky strategy, as some parts can take weeks to arrive.

One solution for that is to go to a two-bin kanban system.
 
Basically you have two bins, and whenever one is empty you open the second one and place a reorder for the empty.  The last company I was with implemented this pretty successfully.  You can have the bin labelled with a reorder card.  When they killed off a bin, they'd pull the label/reorder card, hand it to the buyers, and open the second one.  New stock that came in went into a new bin and was ziptied shut.

You don't painstakingly count each part.  When a bin is opened, it's counted as used and transacted all at once.  Closed bin is still unused.

The key to success is to properly size the bins!  The size per bin depends on lead times of getting replacements and your production levels.  The biggest problem is that if you don't accurately forecast production you end up either with excess stock or, worse, with shortages.  I saw that happen several times during the growing pains of getting it figured out.  It's always a Friday at 5PM when you realize that you're short and need 400pcs on Monday because the next order is stuck on a ship from China for the next 6 weeks.

The lean ideal is to always have just enough and  constantly have more coming in, but the longer lead times (or in other words the inability of that procurement process to scale as you'd want it to) the further you have to get from the ideal and err on the side of too much stock.


When it works, it can work very well, but just be mindful of how your production forecasts are working out.  That balancing game is no different from any other inventory management scheme, though.


Edit:  Actually, I just remembered something kind of interesting in that some suppliers can work the same way.  I've known of contract manucturers that built on a two bin system and had a bin on the shelf waiting for us at all times.  We'd finish a bin, request a new one, they'd send us the box off their shelf and start making a new box worth.  When something would change , like a new update to a product and you anticipate greater production for a while, you'd talk to the vendors and increase bin size accordingly.  Or if you find that you have a pretty available part that sits on the shelf and gets reordered once a year,  you should decrease the bin size to keep your stock moving and only have the excess that you need for a buffer to avoid shortages.

 

[john12ax7]

How do people account for excess needed for production? As inevitably you need more parts due to losses, rework, etc. Do you just slap a fixed percentage on it and revaluate  down the road?

The two bin system seems interesting. With the right sizing would seem more efficient than constantly updating a spread sheet.

 

[audiomixer]

wouldn't it make sense to analyse your past data and test whether you can extrapolate your need in a 3 / 6 / 12 month time frame? what quantities are we talking about anyway? if you can buy 5'000 pcs rolls that might become significant. if you are ordering in tenths it just makes sense to gather some parts to order together, so you can either reduce or eliminate shipping costs. I find that shipping is actually significant part of the costs the way I order.

basically you need a spreadsheet that talks to a agenda / calender. I remember reeding a article about pretty much the same topic, probably on hackaday. small / medium company that hat do scale production wise from garage to 10+ team. maybe from crowdfinding to medium size production. I tried to find the article / video but I had no success. It really had it all - supply chain, manufacture and costs structure, time line, tracking of products and after sale support.... very well done and I remember they had programmes spreadsheets to track inventory, orders, financials, product cycles and to on. they could tell from the shipping date who built it, what revision, what ever. damn! what product was it?

- Michael

 

[Andy Peters]

Hi,

As some of you here run small OEMs, what do you use to keep on top of stock control?

I'm thinking that a tablet PC, running Excel, kept dedicated to the task might work? Every time someone removes or replaces an item of stock they update the tablet (yes - I know this requires discipline...).

In an ideal world, I'd be able to link it with Mouser / Farnell etc and get it to fire off a BOM.

What do you employ for the task?

I have a spreadsheet that acts as a master parts list and inventory control. Each and every part I might want to use has a line in the sheet. It gets assigned a part number, and then I fill in a manufacturer name, manufacturer part number, a description, a quantity-on-hand, an approximate price, and then some other supporting details like the name of the part in the Kicad library.

The schematics contain parts which all have that part number embedded. I've also created some "placeholder" part symbols, which don't have footprints but do have a part number and reference designator. These are parts you want on the BOM for your assembly (screws and other mounting hardware, mostly).

(As a side note, the parts in the parts list and in the Kicad library are "vetted." That means that I've got a symbol in the schematic library, a footprint in the footprint library, and I've verified that I can actually buy the part from distribution. This all gets done once and I never have to worry about it again, unless the part goes obsolete.)

Kicad's EESchema spits out a BOM with the necessary stuff as a CSV file. I export the parts-list spreadsheet to a CSV file too. A Python script parses the two and spits out a proper parts list, with orderable part numbers, quantity and the like. When I pull a kit, I have to make sure that I update all of the quantities manually.

The parts themselves are all in bins which have labels and barcodes.

On the to-do list:

a) update the Python script to tell me how many parts I need to order for a given quantity of boards based on what parts I already have in stock.
b) have the script pull prices directly from Mouser etc.
c) come up with a less-painful way of updating the quantity-on-hand of each part.
d) Replace the spreadsheet with a proper database front-end.

 

[thermionic]

This has been such a useful thread and I am extremely grateful for the generosity of support. Thank you all very much.

There is a lot of food for thought here. Each methodology has attributes that are attractive.

When it comes to 'lean manufacturing' I have learnt the hard way. For example: I revised a product to have less hand wiring, so the parts that had flying leads became mounted on the PCB. However, when it came to build the new version it proved slightly more labour-intensive, as the sequence of events to assemble the new design had to be followed in a specific order, so took longer than it did when flying leads were used... My point is, until you're trialled something you don't know 100% if it's false economy. Often it won't be, but other times...

 

[Rochey]

I'll be doing a live video on my development of all this next week. I initially planned on doing it last night on EA's facebook group.
Given the "uneven" production flow at EA, I need to not only track inventory and have a list of "these are known good parts", but also track each assembled PCB (currently Edens, but also TinyVU's soon), so that I can track them through:

Assembly -> Optical inspection -> Electrical Test/Programming -> Bagging -> Storage -> Shipping.

By uneven, I mean that Assembly could happen one week, then the next step the week after, etc. I need to be able to pick up any PCB and know its status, along with notes. Another example might be that the board failed optical inspection, I made notes on what needed to be fixed, but didn't get to repairing the board until a few months later.

This typically takes an ERP (Enterprise Resource Planning) package to do, but the time to integrate it would kill me. I'm quite happy to keep a "monster sized" spreadsheet for much of this and develop custom front ends for it. (e.g. appsheet is an awesome app for this) along with small python programs for searching and pulling data from the excel file.

My process is to design the label in Zebra label software, then export the script, apply variables, and put the template into a basic python program. It will do both 2D datamatrix barcodes and 1D Code128.
As mentioned, custom python programs will then take those values (e..g PCBA0002) search through and find the right variables and print the label with the data.

Step 1 in lean manufacturing is to document your existing processes. That's what I'm doing at the moment, putting it all into the EA Wiki. (still private). For me, this means documenting the entire manufacturing process, and deciding on label formats etc.

 

[JohnRoberts]

This has been such a useful thread and I am extremely grateful for the generosity of support. Thank you all very much.

There is a lot of food for thought here. Each methodology has attributes that are attractive.

When it comes to 'lean manufacturing' I have learnt the hard way. For example: I revised a product to have less hand wiring, so the parts that had flying leads became mounted on the PCB. However, when it came to build the new version it proved slightly more labour-intensive, as the sequence of events to assemble the new design had to be followed in a specific order, so took longer than it did when flying leads were used... My point is, until you're trialled something you don't know 100% if it's false economy. Often it won't be, but other times...

Back when I was working for a very sharp pencil (cheap) manufacturer, I used to spend a lot of time in the cost engineering department (at least a half dozen people) massaging designs to be more cost effective.  In a larger organization you need to factor in costs besides just direct labor like overhead for different work centers.

The same operation could be more or less expensive in a different factory building due to different overhead.  This led to some perverse outcomes when an expensive work center would get less work run through it making it even more expensive due to fixed costs divided by smaller denominator.

Total cost is what matters and sometimes the total cost is not obvious.

JR