Dick Harris question/answer thread

[PlacaFromHell]

Hi again, Mr. Harris. I know the main purpose of beamspring keycaps is imitate the Selectric family and that you were not part in the design of those, but I wonder if you could confirm if Selectric (and in consequence beamspring) keycaps use curvature continuous tops. That would explain why other spherical keycaps don't look as good.

[zrrion]

springboarding off of that question, the selectric 3 had different keycaps from previous selectrics and that style seems to have set the standard for IBM going forward, was accommodation for that key top style a serious consideration when designing switches?

[Ellipse]

Hi Dick, I just came across your thread and want to echo everyone else in stating it is a great honor that you are a part of our community. Thank you so much for your time reviewing and writing up responses to the community's questions.

I am the coordinator of the project that brought the Keyboard F back into production after many decades and it is a great honor to be able to chat with you and to make for the next generation some more keyboards of the technology that you invented. I and many others consider the Keyboard F (what many of us colloquially call the "Model F") to truly be the best keyboard ever: a great experience to type on and an extremely reliable keyboard that people have used for decades after manufacture - can't say that about too many computer parts made these days!

A question: I might have missed your reply in this thread but did you make adjustments to the buckling spring design based on any sound considerations? Part of the enjoyment, in my opinion, of typing on your invention is the musicality of the key presses and the musical ringing after releasing each pressed key while typing. Was optimizing the sound feedback of similar importance to optimizing key travel distance and tactile feedback in the development / prototyping process?

[Easy_Spinach]

>Dick

[Dick Harris]

Hi again, Mr. Harris. I know the main purpose of beamspring keycaps is imitate the Selectric family and that you were not part in the design of those, but I wonder if you could confirm if Selectric (and in consequence beamspring) keycaps use curvature continuous tops. That would explain why other spherical keycaps don't look as good.

I'm not sure that I totally understand your question, but I'll give it a shot. The spherical keycaps were used on the older Selectric typewriters. The cylindrical keycaps were used on newer Selectrics and the buckling spring keyboards. I think this was an design change driven by appearance. I am not aware of any Human Factors reason for it.

[AnnoyedWalrus]

While I don't have a question, I would like to express my admiration of, first and foremost mr Harris and his inventions but also of the experience and technical knowhow of the members of this forum. I recently joined (after I ordered my repro Model F) and I am stunned by the high level of discussion and debate that I have seen here.

After 25 years on the internet, I'm still in awe by the opportunities it gives.
Thank you and stay safe.

[SneakyRobb]

Hi,

Dick and I will try to upload more soon. He made and sent this video of an original buckling spring prototype/mockup. Hopefully have more info soon.

https://youtu.be/0MaIkV-TV2w

Robb

[thefarside]

Nice video! Looking forward to learning more.

[darkcruix]

This is a wonderful idea - can't wait to see more. Also, I like the shot very much as I have never seen the buckling to the full extent like this ... NICE

[daneli]

Great video! Can we infer that the "click" sound is produced primarily by the spring snapping against the inside of the plastic barrel? (Rather than the flipper touching down?) Also interesting that with this prototype there is a second "click" when the key is released, but little "ping."

[BuGless]

I am stunned by the high level of discussion and debate that I have seen here.
...
After 25 years on the internet, I'm still in awe by the opportunities it gives....

That means you've *just* missed the golden-age of the INTERNET which ended around the beginning of 1994.
I regret to tell you that up till the end of 1993, 99% of all content found on the Internet (which was mostly USENET back in those days) had a comparable level or even higher level of discussion and debate than what can be seen here on DT. There also was NO spam back then (no kidding).

The Internet never recovered from the influx of the masses in the beginning of 1994.

[headphone_jack]

This new video is absolutely golden. What an incredible historical find, and an awesome demonstration of the buckling spring mechanism. Imagine the expense a completely clear buckling spring setup would incur today! Props to Robb for his amazing work with Mr. Harris, without which much of this information would be lost to history.

[vyquad]

if this thread is still active, what happened to the manufacturing equipment used for making model f's and beamsprings?

[SneakyRobb]

Question:
Did you design the spacebar stabilization system of the F XT keyboard? The space bar has an underbarrel stabilization spring that requires disassembly to reinstall. Why was that method chosen over the exterior wire stabilization like on earlier beamsprings 3277, 5251 etc. I know the F122 keyboard and Model M have a more traditional just torsion bar?

Typed answer from Dick Harris:
"I did several space bar designs, but I don't associate the system identifications with the specifics of the designs. My attention was on the mechanism designs and another group configured keyboards for particular systems. If you could sketch or send some pictures of the mechanisms, that would help me respond. For now, I'll tell you what I remember generally about space bar mechanisms. The added weight of the space bar needs to be resisted so that this weight doesn't activate the "space" signal. This requires an additional spring. I designed one stabilizer bar with a slight bend that caused an upward force. Sometimes an additional spring was added if using a normal stabilizer bar. Dust and contamination are always a concern with keyboards so it is desirable to have as many moving parts inside the frame as possible; thus, a stabilizer bar inside the keyboard frame. I've attached a couple publications describing space bar designs. The actual implementation if done could vary considerable from the sketches in either of these concepts."

The PDF was attached to this answer

Space Bar Concepts (1).pdf

(2.33 MiB) Downloaded 220 times

Question:
As well I have noticed the FXT keyboard had "stepped" keys. The later Model M used a 2 unit wide key with 2 barrels, the FXT had a key with 1 barrel with a stepped cap. Even with that 1 barrel though, the Model F design still allowed for good off barrel key presses. Did you design any other method of stabilizing those keys?

Typed answer from Dick Harris:
"Keys that are wider than a single key position require some special attention. The same rotational angles allowed by the key stem guides that don't cause a problem with single key tops, can cause interference with a wider or larger key top. Also, wider key tops can increase the off-center finger forces. This can create a binding condition that causes the key module to fail. This can be more exaggerated as the key module wears. We designed some short stabilizer bars, but I don't remember if they made it into production or not."

[Weezer]

I'd like to know who designed the spacebar stabilizer for the Displaywriter. That's one of the most impressive (as impressive as a stabilizer can be I guess) stabilizers I've seen.

[SneakyRobb]

I'd like to know who designed the spacebar stabilizer for the Displaywriter. That's one of the most impressive (as impressive as a stabilizer can be I guess) stabilizers I've seen.

Hi I discussed with Dick and this should be the production version of the 3rd and 4th page space stab from the pdf above.

[Weezer]

I'd like to know who designed the spacebar stabilizer for the Displaywriter. That's one of the most impressive (as impressive as a stabilizer can be I guess) stabilizers I've seen.

Hi I discussed with Dick and this should be the production version of the 3rd and 4th page space stab from the pdf above.

I saw the illustration but only when going back did I see Dick's name credited. My hat is off to him. Very impressive design there.

[SneakyRobb]

21.
Was there an official IBM method of reattaching a buckling spring that fell out of the module. People now often use a chopstick to press it on. How would you have restuck a spring that came off while the board was assembled?

The way people in production handled this was to use a toothpick or other cylinder object, put the spring into position and push it down from the side.

He was not sure how long they kept doing it, but the original way to attach the spring correctly that quality control actually looked at was where the top of the spring was oriented to the back on the flipper. I made a quick image to show what that means. It might not be intentional but the patent drawing also shows it.

The spring is not ground flat, so it has a slight angle to it where it runs out, so the part where it ends should be at the back. If the spring is installed in a different orientation it would very very slightly alter the actuation. Although likely not noticeable. Not sure how necessary that really was, but that's what he told me they did for at least some time. Maybe someone with an early XT they are taking apart can check.

[pandrew]

Hey Dick,

I have two more questions:

1) Do you happen to know what the purpose of those two floating tiny dots/circles was above the Model F pad card footprint, as shown here? I would guess that they are there for mechanical reasons, perhaps to lift up the flipper pivot point to the same height as the sense copper area, or maybe to ease the pivoting action of the flipper by providing a smoother contact surface with some PCB technology of the time. There have been other non-mechanical guesses too by others.

dots.png (24.66 KiB) Viewed 9367 times

The soviet clone model F had what looks like even bigger circles:

soviet_model_f_circles.png (83.38 KiB) Viewed 9345 times

2) You mentioned before when speaking about Model F keyboards that they are "buckling spring with unconstrained end condition".
Could you clarify what you mean by that? And perhaps what would a constrained end condition buckling spring look like? Wouldn't the way the buckling spring hits the wall of the barrel when it buckles be considered a constrained condition? Maybe I'm just misunderstanding what you're referring to.

[SneakyRobb]

Spoiler:

Hey Dick,

I have two more questions:

1) Do you happen to know what the purpose of those two floating tiny dots/circles was above the Model F pad card footprint, as shown here? I would guess that they are there for mechanical reasons, perhaps to lift up the flipper pivot point to the same height as the sense copper area, or maybe to ease the pivoting action of the flipper by providing a smoother contact surface with some PCB technology of the time. There have been other non-mechanical guesses too by others.
dots.png
The soviet clone model F had what looks like even bigger circles:
soviet_model_f_circles.png

2) You mentioned before when speaking about Model F keyboards that they are "buckling spring with unconstrained end condition".
Could you clarify what you mean by that? And perhaps what would a constrained end condition buckling spring look like? Wouldn't the way the buckling spring hits the wall of the barrel when it buckles be considered a constrained condition? Maybe I'm just misunderstanding what you're referring to.

"1) You are correct. The dots do maintain the relationship of the pivot point and the capacitive sensing surfaces. They eliminate the effects of any variations in copper thickness. The capacitive sensing signal is generally proportional to 1/d where d is the thickness of the gap between the pivot plate (flipper) and the copper surface. With very small gaps signal differences can vary significantly with gap changes. This makes the design very sensitive to dust, particulate contamination and part tolerances that effect this gap. This was likely the main exposure that lead to Keyboard M.

2) Actually the buckling spring is already buckled slightly when the key-stem is assembled. This buckle is the result of the slope of the surface that the top of the spring rests against, and the fact that the spring is rotated before it is pressed onto the flipper so that the spring wire end is oriented at the back (away from the pivot point of the flipper). We discussed this last week. The spring is partially unconstrained at the top to allow the top of the spring to rotate around the front of the top coil. The lower end of the spring is constrained in relation to the flipper, but the flipper is unconstrained just enough to allow the flipper to rotate against the circuit board (capacitor plates). So actually the snap doesn't occur when the spring buckles. It occurs when the force center through the lateral most deflection of the already buckled spring moves across the pivot point of the flipper. To control the key travel required for the system to snap, it is important that the spring is free from any frictional forces caused by rubbing against wall of the barrel. After the snap, it doesn't matter if there is slight friction from the walls of the barrel as long as the spring is free when the key moves upward causing the force center of the spring to move back across the flipper pivot point and resets the key signal. Hope this makes sense.

Just for clarity, the only totally unconstrained spring would be one that is flying slowly through the air. A totally constrained spring would be one that has both ends pressed on a post similar to the post on the flipper, be in some amount of compression and be in a cylinder the diameter of the OD of the spring. Maybe there is still some degree of freedom that isn't constrained here, but you get the idea."

[pandrew]

Thank you Dick,

Enlightening read!

[SneakyRobb]

Spoiler:

Dick,

1. Is it true that the need for a new keyswitch design to follow the beamspring, leading to the buckling spring, was due to a new ISO standard about keyboard height?

2. Are you aware of the modern recreations of the 4704 line of banking terminal Model F keyboards? What do you think of that endeavor?

3. Would you be willing to share info on other prototypes or sketches you may have? They are uniquely fascinating to me and I am certain many others.

"1. Was the choice of the spacing foam in keyboards B and F made with the knowledge that it would degrade over time, before other parts might fail? How might that have been weighted with the use of other materials? That is to say, why was the foam that was used in the end products used, as opposed to other options that might have existed?

It has been a long time, but I'll respond to this with some thoughts and in some cases some assumptions. I'm guessing that the foam used in the beam-spring keyboards and the buckling spring is the same material. If this is correct, the material was originally selected sometime in the early 1970's; and, since the material appeared to be successful on the beam-spring, it should be successful on the buckling spring. Target product life was never expected to be 40 or 50 years so the material seemed a reasonable and economical choice at the time. I don't recall any other material comparison studied evaluating alternate materials. I suspect that a material choice was made that passed our functional tests, met cost constraints and survived our technical reviews at to aging and environmental exposure.

2a. Around what point in the process of designing Keyboard B/beamsprings, was the need for the contamination shield determined? Was it considered at the outset as a necessity, or was it added later on for industrial usage?

We had a spill test requirement that the contamination shield helped pass. I think the contamination shield was included in the first production beam-spring keyboards.

2b.These contamination shields (rubber mats) are almost always decayed and removed in the remaining Keyboard Bs used these days. For regular usage, if kept in a clean environment, does this pose any concerns for the longevity of the keyboard? Or should it be fine to leave them without a shield long term, if the conditions are kept relatively clean?

The contamination shield is intended only for liquid and particulate protection. As long as the user is willing to accept the consequences of potential contamination, the contamination shield is not necessary. If the shield has deteriorated, you are probably better off without it as it will then become a contaminate.

As others have said, thank you again for your time and for answering questions."

You are welcome. These were fun years in the development of technology. Especially for a mechanical engineer.

[Muirium]

Ah, 50 year beamsprings.

Say, what did IBM folk think about Micro Switch and their Hall effect switches? I don’t like them nearly as much as beam or F (especially F) but they do still mechanically work quite nicely; if not the incomprehensible protocols spoken by their PCB mounted controllers. IBM inadvertently made our restorations easier with the modular controller cards, compensating a fair bit for all that crumbly foam!

Also: what did IBM engineers think about Cherry?

[Willy4876]

Dick,

I don't really have a specific question, but I always love listening to other engineers talk about their profession. It's always interesting learning what older generations of engineers have in terms of insights from their careers.

I'm just starting my career as an engineer and I always enjoy listening to older engineers talk. Have you (or perhaps Robb) considered doing a video/interview about your career? I the timeperiod you were present for is very interesting time in technological history and your insights during that period would be very interesting to hear.

Even if you decide not to do this, thank you for sharing your knowledge here. I have greatly enjoyed reading through your comments.

[SneakyRobb]

Ah, 50 year beamsprings.

Say, what did IBM folk think about Micro Switch and their Hall effect switches? I don’t like them nearly as much as beam or F (especially F) but they do still mechanically work quite nicely; if not the incomprehensible protocols spoken by their PCB mounted controllers. IBM inadvertently made our restorations easier with the modular controller cards, compensating a fair bit for all that crumbly foam!

Also: what did IBM engineers think about Cherry?

I asked about keyboards like 3277. He said that IBM had contingencies for anything. In case of the beamspring development failing, the contingency plan was to use microswitch hall effect. For factors, mainly simplicity reliability and similarity microswitch was considered the best non IBM switch. For early systems that would use beamspring keyboards they simultaneously used microswitch sort to see what would happen.

So the opinion of microswitch was extremely high.

He said they tested among others cherry and maxi switch (not sure the timeline for maxi) and at least at that time they did not like them for the reliability criteria, which were extremely high to meet.

The main issue with microswitch was related to lack of click/tactile and dislike of that switch's hysteresis.

[Polecat]

I'm late in finding this very fascinating and informative discussion, but I'm wondering if Mr. Harris has any knowledge of the Model M keyboards with blue Alps SKCM switches that pop up now and then? As far as I know it has never been confirmed whether these were a genuine IBM product, or an unauthorized clone.

[flowerlandfilms]

I'm late in finding this very fascinating and informative discussion, but I'm wondering if Mr. Harris has any knowledge of the Model M keyboards with blue Alps SKCM switches that pop up now and then? As far as I know it has never been confirmed whether these were a genuine IBM product, or an unauthorized clone.

And as an adjunct to this, a general appraisal of Alps switches in general would be fun?

[SneakyRobb]

1
Great video! Can we infer that the "click" sound is produced primarily by the spring snapping against the inside of the plastic barrel? (Rather than the flipper touching down?) Also interesting that with this prototype there is a second "click" when the key is released, but little "ping."

Dick:
The sound on keydepression is a combination of the spring hitting the barrel, ringing afterwards and the flipper hitting the circuit board. The dominate sound comes from the flipper hitting the circuit board. On release, the click is primarily the flipper hitting the housing with a less significant, but audible, contribution from the spring as it suddenly returns to its original nearly straight condition. It's significant in the design that all this sound and tactile sensation is nearly symultaneous and inseparable. I hope this helps.

2
Hi Dick, I just came across your thread and want to echo everyone else in stating it is a great honor that you are a part of our community. Thank you so much for your time reviewing and writing up responses to the community's questions.

I am the coordinator of the project that brought the Keyboard F back into production after many decades and it is a great honor to be able to chat with you and to make for the next generation some more keyboards of the technology that you invented. I and many others consider the Keyboard F (what many of us colloquially call the "Model F") to truly be the best keyboard ever: a great experience to type on and an extremely reliable keyboard that people have used for decades after manufacture - can't say that about too many computer parts made these days!

A question: I might have missed your reply in this thread but did you make adjustments to the buckling spring design based on any sound considerations? Part of the enjoyment, in my opinion, of typing on your invention is the musicality of the key presses and the musical ringing after releasing each pressed key while typing. Was optimizing the sound feedback of similar importance to optimizing key travel distance and tactile feedback in the development / prototyping process?

Dick:
A fundamental objective of the design was to minimize the total height of the keyboard without compromising keyboard performance. Keytravel was the initial parameter used to determine total keyboard height, then keystem bearing system was designed for minimum binding and finally minimum keybutton thickness above the barrel, circuit board and bottom support frame. Total keytravel was determined as 5/32" (4mm). Make/break points were established to ensure that the key always actuated before the key bottomed and reset before the keystem upstop. All this to say that sound was a secondary result of these objectives. In fact, we were concerned that the keyboard was too loud so we constructed a keyboard with strips of foam rubber inside the springs to reduce the springs contribution to the sound and submitted it to Human Factors testing. The results of this test were that there was no significant improvement in performance, and the foam rubber addition was not justified. I just Googled "unintended benefit" and "manna from heaven" was one result. This rung a bell since I'm a follower of Christ and there was much prayer going on during the development process. If credit is due, it goes to the Lord.

3
Hi again, Mr. Harris. I know the main purpose of beamspring keycaps is imitate the Selectric family and that you were not part in the design of those, but I wonder if you could confirm if Selectric (and in consequence beamspring) keycaps use curvature continuous tops. That would explain why other spherical keycaps don't look as good.

Dick:
I'm not sure I fully understand your question, but I'd do my best. Early Selectrics used a spherical keycap top surface while the latest Selectrics used a cylindrical keycap top similar to the one used on Keyboard F. The beam-spring keyboard was a peer to the early Selectrics, and Keyboard F was a peer to the later Selectrics with one difference: all single position keycaps on Keyboard F were the same shape and the keyboard frame was curved to achieve nearly the same "finger reach profile" as the later Selectrics.

4
First Off thanks for starting one of the best made keyboards ever made. Ive never used a model F but i have used both a Unicomp M and a 1993 M

My question is when you first started designing the switch did anyone ever doubt what you were doing?

Dick:
Early prototype keymodules actually worked well. In fact, we were able to make a large scale (8x - I think) model that was a thin slice (slightly thicker than the spring diameter) through the center of the key module. This model was placed on an overhead projector. It was easy to see how it worked, and it instantly revealed the simplicity of the design; however, as we considered manufacturing 1000's of 100+ key keyboards repeatability was a concern. While we could order springs that worked, there was no assurance that the next order would work. We were tasked with proving that the design was repeatable. Our best attempts at dynamic and kinematic analysis at the time were not fruitful so we proceded with parametric studies where spring and parts tolerances were varied, models made and tested and results analysed. The results supported the design. I still think our manufacturing team was more than a little nervous. Production began after some initial hiccups. As far as I know, production proceded smoothly, and I only saw one broked spring from a customer and that failure was the result of an inclusion in the spring wire and had nothing to do with the design.

[del20nd]

Thanks so much for taking the time to answer these questions about your career and designs!

I have a couple questions to add:

About IBM in general: did the culture change noticeably between the beginning and the end end of your career? Do you have any thoughts about the company today that you'd be comfortable with sharing?

On the Model F: you mentioned that the F's design timeline was constrained by the release schedule of the IBM PC. How intertwined were these projects? Was it just a case where you happened to be designing a keyswitch at the right place and the right time which might fit the PC budget, or was the design targeting the home market from the very beginning? How much of the design specs were driven by the fact that IBM was considering shipping it with the PC (cost, key profile, etc?)