Modern way to specify values on schematics

[kuon]

Hello,

I thought I would find tons of resources about that, but surprisingly, I found no recommendations.

If this is a sensible topic, I really don't want to start a flame war, so I hope it won't come to that.

I've seen a lot of way to write values in schematics.

For capacitors I've seen 100nF, 0u1F, 0.1uF or even 0.1µF and the same without the unit (100n, 0u1, 0.1u, 0.1µ).

For resistors I've seen 4M75, 4.75M, 4.75MΩ or 100, 100R, 100Ω.

What would be the best modern practice?

As a side question, should resistors and capacitors always have their value visible? I am using a TI chip (BQ25570), which requires 7 resistors to configure the different voltages and the values are not really helpful at a glance (mega ohms), I am wondering if adding a small note with the resistors values and the formula on the schematics wouldn't be better.

[ataradov]

I have to review schematics a lot from various sources, so I see a variety of this stuff too.

My personal preference is to always use '.' for decimal separator. And don't use use characters that are hard to type and may not be present in all fonts (µ, Ω).

For resistors - drop Ohm in any variation, so 100 (or 100R), 10K. Full spelling takes a lot of space.

Notes are also a good idea, especially for stuff that was calculated and not just taken from the reference. Atmel/Microchip datasheets include calculations for the load capacitors on the crystals. This does not
get in a way, but helpful when you need it.

Other than this - use flexibility and common sense. I'd rather read a schematic by someone who had freedom to express the design.

One thing that I really like about "modern" designs is lack of standards. This may sound strange, but when you are confined to IEC, GOST, whatever, you sooner or later end up with a mess. Those standards don't deal with complex designs very well.

[Benta]

For a schematic, I only use the reference, eg, R1, R5, C2, U3 etc.
For a PCB layout I do the same thing.
I place the component values in the "mouse-over" component description / note / comment  field in the design files (schematic and PCB both).

This gives you a nice clean schematic that's easy to read, and with a click or mouse-over on the component provides additional information.

Full information (such as supplier etc.) is in the BOM.

[MarkF]

For a schematic, I only use the reference, eg, R1, R5, C2, U3 etc.
For a PCB layout I do the same thing.
I place the component values in the "mouse-over" component description / note / comment  field in the design files (schematic and PCB both).

This gives you a nice clean schematic that's easy to read, and with a click or mouse-over on the component provides additional information.

Full information (such as supplier etc.) is in the BOM.

My paper doesn't have a mouse. 
I like to see the value on the schematic.

And do not use 'm' for micro.  I use 'm' for milli and 'u' for micro.

[Gary350z]

I too would like to see the answers to this question.

All through my schooling and work, I've always seen 4.75M. Just since I've been on this forum I've seen the 4M75. Dave has done a video or two on this, and has said that the deciamal point in 4.75M may be become unreadable on a paper schematic, a good point.

I'm in the USA, is it different in other countries.

[robertbaruch]

I follow a few rules:

1. Never use a decimal point. Sometimes it doesn't show up, and is easily missed. In place of the decimal point, use the magnitude. Try to avoid leading zeros if possible.

BAD: 1.2k
GOOD: 1k2
BAD: 0.1u
GOOD: 0u1
BETTER: 100n

2. Never use the unit, only the magnitude, EXCEPT when the magnitude is unity. This is just practical, because you really don't want to use any more space than you have to. The schematic symbol itself shows you what the component is. Resistances are always in ohms, capacitances in Farads, inductances in Henries. For ohms, use "R".

Corollary to #2: I prefer to use the IEC symbol for resistors, which is a rectangle. The value goes inside the rectangle instead of next to the symbol.

BAD: 1kΩ
GOOD: 1k
BAD: 1.2Ω
GOOD: 1R2
BAD: 22pF
GOOD: 22p

3. Never use the micro character. Instead, use "u".

BAD: 22µ
GOOD: 22u


As for the question about resistor values, the primary purpose of a schematic is to convey information, so it needs to be as visually pleasant as possible. Lines which cross every which way are bad. Even multiple parallel lines for a bus isn't great, use a thicker bus line instead. Signals should generally flow from left to right. Use subsystems if things get too busy.

So, if adding component values would clutter up the diagram, it's reasonable to put them in a table off in the corner. Use notes to explain the values.

[ataradov]

has said that the deciamal point in 4.75M may be become unreadable on a paper schematic, a good point.

This is a problem when you rely on the paper schematic for everything. But your manufacturing data is digital. So at most you are in for a minor confusion if you miss the point.

For me personally reading "M"  in the middle of the digits is harder.

[Benta]

My paper doesn't have a mouse. 
I like to see the value on the schematic.

And do not use 'm' for micro.  I use 'm' for milli and 'u' for micro.

I'm 100% with you, I do design on paper as well and note the values directly there.
But when you go from (verified) design to documentation the situation is different. Then it's about clarity and readability, and I try to avoid cluttering the schematics with unnecessary stuff. That's why I use the approach with underlying information I described above.

Cheers.

[rstofer]

Side issue:  When I look at an op amp circuit, the first thing I try to do is estimate the gain.  I don't have to get it down to 3 decimal places but I would like the order of magnitude.  Same story for integrators, what's the time constant?  I can't reasonably do this if I have to hunt for values in a table.

[dmills]

There is a certain generation of engineer who grew up with third generation photocopied diagrams from before photocopiers were digital, we tend to avoid decimal points like the plague that they were....

For me, everything gets a unit, and if the unit is unity it gets an R or such, that way there is never any ambiguity about "Is a resistor labelled simply 100, 100 ohms or 100k, 100R is 100 ohms, 100k is 100k ohms, 0R1 is 0.1 ohms.

One other note, enough with hiding nets already, if there are 27 ground pins, they should be on the schematic somewhere (Probably in a separate part symbol, but they need to be there), same with power pins, don't put them in the main bit of the diagram (unless that makes sense) but do show them somewhere (FPGA folk looking at you, and yes I know there are like 300 power pins on 6 different rails....). It can give the layout guy some hints about how you want the decoupling done apart from anything else.

I would usually put values on the schematic, sometimes very hard to tell the difference between a coupling cap and a loop stability cap if you cannot see that one is hundreds of nF and the other is 10pF, the one exception might be decoupling something like an FPGA but it would get a large note "All non polar capacitors on this sheet are 10nF 0402 X7R". 

Ultimately the really important thing is that the schematic flows well and makes it easy for the reader to identify the functional blocks and expected behaviours. On that note, labelling things like regulator outputs with the expected range of voltages is a good thing (3.25 - 3.36V) for example, and label the test points (On the board if there is anything like sufficient space).

There are two really nice things about PDF, you can have extra pages at no cost, so there is no need to shoe horn way too much stuff onto a page (Japanese radio manufacturers looking at you), and you can have hover over links to details and even data sheets.

Regards, Dan.

[Gyro]

It's best to stick with engineering number format too, ie. powers of 3. That way you never see 0.1u, it's always 100n.

Personally, I much prefer 4k7 over 4.7k, It's less susceptible to lost dots and bad eyesight. I use the same in speech too.

[T3sl4co1l]

There may be some confusion over m/M.  Meg is preferred in that case.

The typical example is SPICE, which is case insensitive, so 'M' always means milli.  'MEG' means mega, or you can use a multiplier instead ('1e6').  If you're doing a full design (schematic capture, simulation and BOM) you'll probably need to do this.

Tim

[AndyC_772]

For a schematic, I only use the reference, eg, R1, R5, C2, U3 etc.
For a PCB layout I do the same thing.
I place the component values in the "mouse-over" component description / note / comment  field in the design files (schematic and PCB both).

This gives you a nice clean schematic that's easy to read, and with a click or mouse-over on the component provides additional information.

Full information (such as supplier etc.) is in the BOM.

Component values are essential information, they need to be shown always. Missing them out doesn't remove clutter, it obfuscates the function of the circuit.

Hidden power pins are the same: just say 'no'.

I tend to relegate the tolerance and dielectric to the BoM, but even they're shown on the schematic if they're at all unusual.

Ask yourself this: does a printout of this schematic tell me everything I need to debug the circuit?

[T3sl4co1l]

A catch about power pins: in some tools, these connect or hide automatically (between different parts of a multipart component).  Others, you have to hide them manually.  Or there's the method of using a separate power supply part.

In any case, don't scatter multipart components across multiple sheets, keep them local, and place the supplies nearby.

Tim

[dmills]

Good advice until you hit the 600+ ball FPGA sort of part, where a few sheets for the power, one for the DDR IO bank, one for the network, a couple for the serdes starts to make a lot of sense.

You really cannot do a sane schematic with a part with that many connections without carving the thing up into multiple IO banks so that the schematic breaks down by function.

Regards, Dan.

[CatalinaWOW]

Atarodov hit the key point.  Make the schematic usable.  If that requires violating rules fine. 

Make the paper schematic usable.  While the designer and manufacturer may always have access to the electronic version, the field tech or the guy who bought your stuff on the used market 20 years after sale probably won't.  And even for the original designer paper often proves useful.  I have yet to see the operation that can afford more monitors than sheets of paper and it is often necessary to refer to more than one schematic at once, and not infrequent to need several.

I'm torn on the number format.  For my generation the 4R5 format is foreign and requires mental gear shifting.  But it is the format that will be printed on the part if anything is printed at all.

[bsfeechannel]

The SI prefixes were invented to avoid decimal points and commas. So, in theory you should write: 470mΩ, instead of 0.47Ω, or 2200pF, instead of 2.2nF. In practice, decimal points are really replaced by the prefixes but not exactly as dictated by the SI. When the unit is obvious it is left out.

I prefer 0R47, 100, 1k, 2k2, etc. for resistors, 2p2, 330p, 4n7, 68n, 2µ2, 10µ, 3300µ for capacitors, and 150n, 2µ2, 330µ, 15m, 10H for inductors.

There are traditional scripts, like 0.01mmF or 10kpF for 10nF, that you can find in vintage schematics. But they are a thing of the past, I guess.

[AndyC_772]

Good advice until you hit the 600+ ball FPGA sort of part, where a few sheets for the power, one for the DDR IO bank, one for the network, a couple for the serdes starts to make a lot of sense.

Absolutely right. Separate the power pins out into a separate symbol, and put them on a different page, but don't actually hide them completely.

For some reason, when creating symbols in OrCAD, the option to completely hide any pin designated as a power pin is actually selected by default 

[bson]

Once upon a time people made photocopies of schematics.  Things like connection dots and periods in numbers risked being dropped by an unfortunate blank spot in the copy.  By using notation like 2R2 consistently there was never a question of if that speck was a mangled period or just a bit of dirt.

We no longer do that; if I provide you with a schematic (it won't be on paper) and you worry about losing periods or connection dots in photocopies you're doing it wrong.

TL;DR: it's a workaround for a problem that no longer exists.

[filssavi]

Not related to component values but worth noting nonetheless...

When doing the schematic alerts put the actual nominal voltage in the name of power rail nets so when a different guy will have to troubleshoot your multi board design won’t have to go spelunking around trying to understand what the actual f**king voltage VDD, VDDA, VCC,VCCEXT etc are.

Also I never have seen in the 5 years I’m in the field someone photocopying a schematic, so I would say that decimal points are fine (If your printer has trouble printing dots, it has done its job and it’s fine to replace it)

[bsfeechannel]

Once upon a time people made photocopies of schematics.  Things like connection dots and periods in numbers risked being dropped by an unfortunate blank spot in the copy.  By using notation like 2R2 consistently there was never a question of if that speck was a mangled period or just a bit of dirt.

We no longer do that; if I provide you with a schematic (it won't be on paper) and you worry about losing periods or connection dots in photocopies you're doing it wrong.

TL;DR: it's a workaround for a problem that no longer exists.

I don't think the problem has to do with printed or photocopied schematics. The way decimal separators and digit groupings are represented is different from country to country. So 123,456,789.0123 can be variously represented as

123.456.789,0123
123 456 789,0123
123'456'789.0123
123·456·789,0123
12,34,56,789.0123

depending where you are.

Using the prefixes leaves no margin for ambiguities.

[filssavi]

Once upon a time people made photocopies of schematics.  Things like connection dots and periods in numbers risked being dropped by an unfortunate blank spot in the copy.  By using notation like 2R2 consistently there was never a question of if that speck was a mangled period or just a bit of dirt.

We no longer do that; if I provide you with a schematic (it won't be on paper) and you worry about losing periods or connection dots in photocopies you're doing it wrong.

TL;DR: it's a workaround for a problem that no longer exists.

I don't think the problem has to do with printed or photocopied schematics. The way decimal separators and digit groupings are represented is different from country to country. So 123,456,789.0123 can be variously represented as

123.456.789,0123
123 456 789,0123
123'456'789.0123
123·456·789,0123
12,34,56,789.0123

depending where you are.

Using the prefixes leaves no margin for ambiguities.

why would you need so many digits though that is the whole reason to use SI prefixes
Also are you really specifying 1Mohm resostors(for example) down to the single ohm? That seems a bit unnecessary

[Wimberleytech]

Once upon a time people made photocopies of schematics.  Things like connection dots and periods in numbers risked being dropped by an unfortunate blank spot in the copy.  By using notation like 2R2 consistently there was never a question of if that speck was a mangled period or just a bit of dirt.

We no longer do that; if I provide you with a schematic (it won't be on paper) and you worry about losing periods or connection dots in photocopies you're doing it wrong.

TL;DR: it's a workaround for a problem that no longer exists.

I don't think the problem has to do with printed or photocopied schematics. The way decimal separators and digit groupings are represented is different from country to country. So 123,456,789.0123 can be variously represented as

123.456.789,0123
123 456 789,0123
123'456'789.0123
123·456·789,0123
12,34,56,789.0123

depending where you are.

Using the prefixes leaves no margin for ambiguities.

I am a Texan (Remember the Alamo!), so it pains me to say this...Europeans got it right with the metric system and with the comma.  It is the right thing to do. 

[bsfeechannel]

why would you need so many digits though that is the whole reason to use SI prefixes
Also are you really specifying 1Mohm resostors(for example) down to the single ohm? That seems a bit unnecessary

This is just to show how the use of decimal separators and digit grouping can be confusing.

I am a Texan (Remember the Alamo!), so it pains me to say this...Europeans got it right with the metric system and with the comma.  It is the right thing to do. 

Howdy partner. Welcome to earth. A bizarre place where we can't even agree where to put our commas.

[emece67]

.

[ebastler]

I prefer 0R47, 100, 1k, 2k2, etc. for resistors, 2p2, 330p, 4n7, 68n, 2µ2, 10µ, 3300µ for capacitors, and 150n, 2µ2, 330µ, 15m, 10H for inductors.

The "10H" is the odd one out in that list. Any particular reason why you include the unit there, and only there?

[AndyC_772]

I think the inconsistency there is that "100" (resistance) has no unit; I'd use 100R.

[Zero999]

I am a Texan (Remember the Alamo!), so it pains me to say this...Europeans got it right with the metric system and with the comma.  It is the right thing to do. 

I agree about the metric system but using commas vs full stops for decimal places seems arbitrary. In the UK we use the . for the decimal place and , for the separator, which only looks right to me because it's what I'm accustomed to. I think the . or , is fine for the decimal place but to avoid confusion only the space or ' should be used for the separator.

And do not use 'm' for micro.  I use 'm' for milli and 'u' for micro.

I tend to avoid using milli farads for that reason, so 1mF should be written as 1000µ. For larger values it starts for make sense to use farads, so 10 000µF becomes 0.01F, which in theory could be 0F01, but that would confuse lots of people.

[jpanhalt]

Interesting how so many feel they are using the SI system, often called the metric system, but don't follow its rules.

Source: https://www.nist.gov/pml/weights-and-measures/writing-metric-units
Rule(Spacing):
"A space is used between the number and the symbol to which it refers. For example: 7 m, 31.4 kg, 37 °C."

In any event, as for schematics or any published work, it is far more productive just to get the "Instructions for Authors" and follow them than to try to argue with an editor.

Such cultural things as 9R1 versus 1n0 seem inconsistent, but common.

[ebastler]

Interesting how so many feel they are using the SI system, often called the metric system, but don't follow its rules.

You are confusing the SI system of units with the typesetting rules for writing numbers and (whichever) units.

[jpanhalt]

Interesting how so many feel they are using the SI system, often called the metric system, but don't follow its rules.

You are confusing the SI system of units with the typesetting rules for writing numbers and (whichever) units.

I wrote SI "system."   That system defines printing format among many other things.  No confusion on my part.  Here is a link to a primary source: https://www.bipm.org/en/publications/si-brochure/

[bitbanger]

This somewhat humorously reminds me I once use the R separator in a project (i.e. 47R) I was worried about distiller/render compatibility issues and the ohm symbol. My supervisor at the time lost her mind over it, "we just don't do that here".   

[Wimberleytech]

I follow a few rules:

1. Never use a decimal point. Sometimes it doesn't show up, and is easily missed. In place of the decimal point, use the magnitude. Try to avoid leading zeros if possible.

BAD: 1.2k
GOOD: 1k2
BAD: 0.1u
GOOD: 0u1
BETTER: 100n

2. Never use the unit, only the magnitude, EXCEPT when the magnitude is unity. This is just practical, because you really don't want to use any more space than you have to. The schematic symbol itself shows you what the component is. Resistances are always in ohms, capacitances in Farads, inductances in Henries. For ohms, use "R".

Corollary to #2: I prefer to use the IEC symbol for resistors, which is a rectangle. The value goes inside the rectangle instead of next to the symbol.

BAD: 1kΩ
GOOD: 1k
BAD: 1.2Ω
GOOD: 1R2
BAD: 22pF
GOOD: 22p

3. Never use the micro character. Instead, use "u".

BAD: 22µ
GOOD: 22u


As for the question about resistor values, the primary purpose of a schematic is to convey information, so it needs to be as visually pleasant as possible. Lines which cross every which way are bad. Even multiple parallel lines for a bus isn't great, use a thicker bus line instead. Signals should generally flow from left to right. Use subsystems if things get too busy.

So, if adding component values would clutter up the diagram, it's reasonable to put them in a table off in the corner. Use notes to explain the values.

Ditto...and NEVER USE FOUR-WAY CONNECTIONS.  NEVER USE DOTS to indicate a connection.

[ebastler]

I wrote SI "system."   That system defines printing format among many other things.  No confusion on my part.  Here is a link to a primary source: https://www.bipm.org/en/publications/si-brochure/

Thanks, I learned something new!
I will limit my claim to "I am using SI units" going forward. 

[ebastler]

NEVER USE FOUR-WAY CONNECTIONS.  NEVER USE DOTS to indicate a connection.

I agree with the "no four-way junctions". But for three-way junctions, don't pretty much all CAD systems use dots to indicate junctions, and don't DIN, IEC etc. require them in schematics?

[Richard Crowley]

Corollary to #2: I prefer to use the IEC symbol for resistors, which is a rectangle. The value goes inside the rectangle instead of next to the symbol.

The problem with that rectangle is that it is confusingly overloaded. If ONLY resistors used the rectangle, OK. But there appear to be half a dozen different components that use the same rectangle (or a slight variation).  Maybe it is a resistor, but there are some places where maybe it is a fuse.  Or an inductor,  or ?

I grew up reading the schematics in Popular Electronics here in the US and that was always my reference for nice design.

[Zero999]

Interesting how so many feel they are using the SI system, often called the metric system, but don't follow its rules.

Source: https://www.nist.gov/pml/weights-and-measures/writing-metric-units
Rule(Spacing):
"A space is used between the number and the symbol to which it refers. For example: 7 m, 31.4 kg, 37 °C."

Probably because it's a stupid rule.

There are also other stupid things such as kg for the base measurement of mass and µ, for micro, which isn't found in all fonts, but u is generally accepted because it's close enough.

[level6]

NEVER USE FOUR-WAY CONNECTIONS.  NEVER USE DOTS to indicate a connection.

I agree with the "no four-way junctions". But for three-way junctions, don't pretty much all CAD systems use dots to indicate junctions, and don't DIN, IEC etc. require them in schematics?

I also thought dots are required for junctions  (not talking about 4 way junctions). Eagle adds them in by default when connecting nets.

[Wimberleytech]

NEVER USE FOUR-WAY CONNECTIONS.  NEVER USE DOTS to indicate a connection.

I agree with the "no four-way junctions". But for three-way junctions, don't pretty much all CAD systems use dots to indicate junctions, and don't DIN, IEC etc. require them in schematics?

I also thought dots are required for junctions. Eagle adds them in by default when connecting nets.

As long as four-way connections are NOT allowed, then dots are redundant.  I am ok with a redundant dot...a pacifier dot.  CAD systems are self checking so they won't make errors confusing crossovers with connections.  But drafted schematics are notorious for this problem.  I have reviewed many thousands of schematics in journal papers and patents.  I see this issue all the time.  Been awhile since I drew a schematic in Cadence, but pretty sure that tool does not allow four-way connections.  Cannot speak for Eagle.

[kuon]

Thank you so much for all your insight.

I have switched to this system https://www.eevblog.com/forum/beginners/modern-way-to-specify-values-on-schematics/msg2238423/#msg2238423 . The arguments makes the most sense to me and I can understand it.

[bson]

Use either period or comma, it doesn't matter which.  Use SI prefixes for 1000s.  Don't use the metric 10 and 100 prefixes (centi, deci etc).  Don't use a space or any other separator for 1000s, keep it simple.  Spell out units other than Ω, simply because some software still has problems with UTF.  Or, use it if you like - it's rarely a problem.  Or skip the units altogether if it's obvious from the component.  It doesn't terribly matter.  The SI m means milli and nothing else.  If you're concerned with UTF and reproducing µ, then use u.

[bsfeechannel]

I prefer 0R47, 100, 1k, 2k2, etc. for resistors, 2p2, 330p, 4n7, 68n, 2µ2, 10µ, 3300µ for capacitors, and 150n, 2µ2, 330µ, 15m, 10H for inductors.

The "10H" is the odd one out in that list. Any particular reason why you include the unit there, and only there?

Because in some old schematics, inductors with unit values would be considered nH or µH.